faculty of health and medical sciences...to my host of co-supervisors marie louise de bruin,...
TRANSCRIPT
U N I V E R S I T Y O F C O P E N H A G E N F A C U L T Y O F H E A L T H A N D M E D I C A L S C I E N C E S
PhD Thesis Asbjørn Nøhr-Nielsen
Pharmacokinetic-Pharmacodynamic Modelling of Fixed-Dose Combinations: A Regulatory Perspective
Academic advisor: Trine Meldgaard Lund
This thesis has been submitted to the Graduate School of Health and Medical Sciences, University of
Copenhagen: 24. April 2020
P a g e | 2
Preface This thesis was submitted as part of the requirements for attaining the PhD degree at the Faculty of
Health and Medical Sciences, University of Copenhagen.
The thesis is based on work carried out in the period from January 2017 to December 2019.
The PhD project is part of the Copenhagen Centre for Regulatory Science (CORS) located at the
University of Copenhagen. The research projects conducted at CORS are focused on “The science of
developing new tools, standards, and approaches to evaluate the efficacy, safety, quality and
performance of medical products in order to assess benefit-risk and facilitate a sound and
transparent regulatory decision making.” The project is one of several that are supported by CORS
and the collaborative companies: Novo Nordisk, Ferring Pharmaceuticals, H. Lundbeck, and LEO
Pharma as well as the patient organization Rare Diseases Denmark.
During the project, I was enrolled at the Department of Drug Design and Pharmacology, Faculty of
Health and Medical Sciences.
This thesis is based on the following papers:
Nøhr‐Nielsen A, De Bruin ML, Thomsen M, Pipper CB, Lange T, Bjerrum OJ, Lund TM. Body of
evidence and approaches applied in the clinical development program of fixed‐dose combinations
in the European Union from 2010‐2016. Br J Clin Pharmacol. 2019;(July 2018):1–12.
Nøhr‐Nielsen A, Lange T, Forman JL, Papathanasiou T, Foster DJR, Upton RN, Bjerrum OJ, Lund TM.
Demonstrating Contribution of Components of Fixed-Dose Drug Combinations Through Longitudinal
Exposure-Response Analysis. AAPS J 2020 222 22:1–14.
Nøhr‐Nielsen A, Bagger SO, Brünner N, Stenvang J, Lund TM. Pharmacodynamic modelling reveals
synergistic interaction between docetaxel and SCO-101 in a docetaxel-resistant triple negative
breast cancer cell line. Eur J Pharm Sci 2020 105315.
P a g e | 3
Supervisors
Principal supervisor:
Associate Professor Trine Meldgaard Lund
Department of Drug Design and Pharmacology
Faculty of Health and Medical Sciences
University of Copenhagen, Denmark
Co-supervisors: Professor Marie Louise De Bruin
Copenhagen Centre for Regulatory Science
University of Copenhagen, Denmark
Professor emeritus Ole Jannik Bjerrum
Department of Drug Design and Pharmacology
Faculty of Health and Medical Sciences
University of Copenhagen, Denmark
Associate professor Theis Lange
Department of Public Health
Faculty of Health and Medical Sciences
University of Copenhagen, Denmark
PhD Mikael Thomsen
Conrig Pharma ApS
Ole Maaløes Vej 3, Copenhagen
Assessment Committee Associate Professor Uffe Kristiansen (Chair)
Department of Drug Design and Pharmacology
University of Copenhagen, Denmark
Professor Janet Mifsud (PhD)
Clinical Pharmacology & Therapeutics
University of Malta, Malta
Clinical Associate Professor Troels Bergmann (PhD, MD)
Dept. Clinical Biochemistry and Pharmacology
University of Southern Denmark, Denmark
P a g e | 4
Acknowledgments First and foremost, a sincere thank you is owed to my academic advisor Trine Meldgaard Lund for
giving me the opportunity to work on a project I find deeply interesting and for her support during
the endeavor. Likewise, a thank you to Ole Jannik Bjerrum for ensuring that there never was a lack
of enthusiasm surrounding the project and for overall encouragement.
To my host of co-supervisors Marie Louise de Bruin, Christian Pipper, Theis Lange, Mikael Thomsen,
I want to express my appreciation for the time you took to assist me during the various projects and
for the discussions we had along the way. To Julie Lyng Forman, I want to underline how important
her assistance and input was for me during my project. To the Copenhagen Centre for Regulatory
Science, I want to express my gratitude for the financial support as well as the discussions and events
that took place during the project.
I owe a special thank you to David Foster and Richard Upton for having me in their group when I
visited the University of South Australia in Adelaide. I also want to thank them, Jim, and Ahmad, for
creating a fun and welcoming environment when I was there.
For an office with a great atmosphere, particularly when everybody is there, Eva, Mia, Yassine, and
Theo is owed a thanks, they made sure it was fun to be part of the group.
A thank you to my friends and family for the patience with me during the project, especially during
the last part. And finally, to Katrine, I couldn’t have done it without your support.
Asbjørn Nøhr-Nielsen
April 2020 Copenhagen, Denmark
P a g e | 5
Abbreviations AIC Akaike information criterion
ATC Anatomical therapeutic chemical
AUC Area under the curve
BIC Bayesian information criterion
CL Clearance
CWRES Conditional weighted residuals
EMA European Medicines Agency
EPAR European Public Assessment Report
FDA Food and Drug Administration
FEC Free equivalent combination
GPDI General pharmacodynamic interaction model
HAART Highly Active Antiviral Retrovirus Therapy
HIV Human immunodeficiency virus
ICH International Council for Harmonisation of Technical Requirements for
Pharmaceuticals for Human Use
IVIVE In vitro in vivo extrapolation
LADME Liberation; Absorption; Distribution; Metabolism; Excretion
OFV Objective function value
PD Pharmacodynamics
PK Pharmacokinetics
PK-PD Pharmacokinetic-Pharmacodynamic
T2DM Type 2 diabetes mellitus
TNBC Triple-negative breast cancer
V Volume of distribution
VPC Visual predictive check
WHO World Health Organization
P a g e | 6
Glossary Adherence Filling or refilling drug prescriptions on time
Compliance Taking medication on time
Drug development tool Tools, such as methods, materials, or measures that facilitate drug
development
Free equivalent combination Separate components of a corresponding fixed-dose combination
Pharmacodynamics The relationship between the concentration of a pharmaceutical drug
and the effect of that drug
Pharmacokinetics The liberation, absorption, distribution, metabolism, and excretion of
pharmaceutical drugs
Substance status The approval state of each component in a fixed-dose combination
P a g e | 7
Summary (English) The use of combination drug therapies is central to the successful treatment of several diseases,
where monotherapies are not efficacious enough or where resistance to treatment emerges. The
development of new drug combinations is, therefore, a major focus. This is also true for fixed-dose
combinations, where there has been a rise in approvals in recent years. Developing fixed-dose
combinations often involves conducting large factorial design studies to verify the efficacy of the
combination. With a greater focus on the personalization of medicines, several dose levels of fixed-
dose combinations will need to be available for patients. For a factorial design study, this will result
in very costly clinical trials. In order to keep developmental costs low and guide drug development,
the validation of existing tools, and the development of new tools is necessary. Such model-based
tools used for the analysis of fixed-dose combinations are, however, only in their infancy.
Therefore, it was the overall aim of this PhD thesis to explore, develop, and validate the use of
modeling tools for the development of fixed-dose combinations. The conducted research in this
thesis resulted in three publications, which addressed the overall aim and objectives of the thesis.
In the first paper, the underuse of model-based approaches in the development of fixed-dose
combinations was identified for the fixed-dose combinations approved by the European Medicines
Agency (58% of approvals). Additionally, although interesting strategies were employed to reprofile
drugs and utilize prior knowledge to avoid dose-finding trials, very few fixed-dose combinations
were geared towards personalization or could be considered innovative. Furthermore, the
importance of pharmacokinetic modeling in selecting the correct doses was highlighted in the study.
In the second paper, clinical trial simulations were performed to assess the feasibility of performing
longitudinal exposure-response modeling of fixed-dose combinations. In a previous study,
exposure-response analysis of fixed-dose combinations had been shown to cause an inflated false
positive rate. In the present research, this was avoided by employing longitudinal exposure-
response to analyze fixed-dose combinations. Thus, longitudinal exposure-response analysis
constitutes a new method for fulfilling the regulatory requirement of demonstrating each
component's contribution to the overall effect.
In the third paper, combination models were evaluated in a preclinical breast cancer study. A novel
combination was analyzed for its potential pharmacodynamic interaction by employing the general
pharmacodynamic interaction model. Based on this model, the combination was characterized as
synergistic with one compound increasing the potency of the other by up to 60% when administered
together. Additionally, analysis of the model provided insights into optimal dose ratios, which can
be used to guide further investigations of the combination.
In conclusion, this PhD thesis explored the current state of model-based development of fixed-dose
combinations, developed a new methodology for the evaluation of fixed-dose combinations, and
validated existing tools for the analysis of fixed-dose combinations. Based on the research
conducted in this thesis, the overall recommendation is that modeling tools should to a greater
extent be incorporated in the development of fixed-dose combinations as they assist in determining
efficacy and provide valuable information to guide development.
P a g e | 8
Summary (Danish) Brugen af kombinationsbehandling er central for den vellykkede behandling af flere sygdomme,
hvor monoterapier ikke er effektive nok, eller resistens overfor behandling opstår. Udviklingen af
nye lægemiddelkombinationer er derfor et stort fokus. Dette gælder også for fixed-dose
combinations, hvor der har været en stigning i godkendelser i de senere år. Udvikling af fixed-dose
combinations involverer ofte udførelse af store faktorielle kliniske studier for at verificere
kombinationens effektivitet. Med et større fokus på personalisering af medicin, skal flere
dosisniveauer af fixed-dose combinations være tilgængelige for patienter. For et faktorielt design
vil dette resultere i meget dyre kliniske forsøg. For at holde udviklingsomkostningerne lave og
vejlede lægemiddeludvikling, er validering af eksisterende og udvikling af nye værktøjer nødvendig.
Sådanne modelbaserede værktøjer, der bruges til analyse af fixed-dose combinations, er imidlertid
kun i deres begyndelse.
Det var derfor det overordnede mål med denne ph.d. afhandling at undersøge, udvikle og validere
brugen af modelleringsværktøjer til udvikling af fixed-dose combinations. Forskningen i denne
afhandling resulterede i tre publikationer, som opfylder afhandlingens overordnede mål og delmål.
I den første publikation blev manglende brug af modelbaserede tilgange til udvikling af fixed-dose
combinations identificeret for fixed-dose combinations godkendt af European Medicines Agency
(58% af godkendelser). Det blev vist, at selvom der blev anvendt interessante strategier til at
reprofilere lægemidler og anvende forudgående viden for at undgå dose-finding, var meget få fixed-
dose combinations rettet mod personalisering eller kunne betragtes som innovative. Desuden blev
vigtigheden af farmakokinetisk modellering ved valg af de korrekte doser understreget i studiet.
I den anden publikation, blev simulering af et klinisk studie udført for at vurdere muligheden for at
udføre longitudinal exposure-response modellering af fixed-dose combinations. Exposure-response
analyse af fixed-dose combinations har i et tidligere studie vist sig at forårsage en forhøjet falsk
positiv rate. Forskningen i nærværende studie viste at longitudinal exposure-response var i stand til
at analysere fixed-dose combinations uden den forhøjede falsk positive rate, der var set i det
tidligere studie. Longitudinal exposure-response giver derfor en ny metode til at opfylde det
regulatorisk krav om at demonstrere hver enkelt komponents bidrag til den samlede effekt.
I den tredje publikation blev kombinationsmodeller evalueret i en præklinisk
brystkræftundersøgelse. En ny kombination blev analyseret for dens potentielle farmakodynamiske
interaktion ved anvendelse af the general pharmacodynamic interaction model. Baseret på denne
model blev kombinationen karakteriseret som synergistisk hvor den ene komponent øgede den
andens styrke med op til 60%, når de blev administreret sammen. Derudover gav analyse af
modellen indsigt i optimale dosisforhold, som kan bruges til at guide yderligere undersøgelser af
kombinationen.
Denne ph.d. afhandlingen undersøgte den aktuelle tilstand af modelbaseret udvikling af fixed-dose
combinations, udviklede en ny metode til evaluering af fixed-dose combinations og validerede
eksisterende værktøjer til analyse af fixed-dose combinations. Baseret på den forskning, der er
foretaget i denne afhandling, er den overordnede anbefaling, at modelleringsværktøjer i højere grad
skal indarbejdes i udviklingen af fixed-dose combinations, da de hjælper med at bestemme effekt
og give værdifulde oplysninger til at guide udviklingen.
P a g e | 9
Table of Contents Preface ................................................................................................................................................. 1
Supervisors ........................................................................................................................................... 3
Assessment Committee ....................................................................................................................... 3
Acknowledgments ................................................................................................................................ 4
Abbreviations ....................................................................................................................................... 5
Glossary ................................................................................................................................................ 6
Summary (English) ............................................................................................................................... 7
Summary (Danish) ................................................................................................................................ 8
Table of Figures .................................................................................................................................. 11
1 Aim & Objectives .......................................................................................................................... 12
1.1 Objectives ............................................................................................................................ 12
2 Introduction .................................................................................................................................. 13
2.1 Fixed-dose combinations .................................................................................................... 15
2.1.1 Combination therapy ..................................................................................................... 15
2.1.2 Fixed-dose combination background ............................................................................. 16
2.1.3 Advantages of fixed-dose combination ......................................................................... 17
2.1.4 Development of Fixed-dose combinations .................................................................... 19
2.1.5 European Public Assessment Reports ............................................................................ 22
2.2 Combination therapy in diabetes & breast cancer ............................................................. 23
2.2.1 Type 2 diabetes mellitus ................................................................................................ 23
2.2.2 Triple-negative breast cancer ......................................................................................... 24
2.3 Pharmacokinetic-Pharmacodynamic Modelling & Simulation ........................................... 26
2.3.1 Background ..................................................................................................................... 26
2.3.2 Population PK-PD modeling ........................................................................................... 26
2.3.3 Exposure-response analysis ........................................................................................... 29
2.3.4 Combination modeling ................................................................................................... 29
3 Summary of results....................................................................................................................... 32
3.1 Paper 1 - Body of evidence and approaches applied in the clinical development program of
fixed‐dose combinations in the European Union from 2010‐2016 ............................................... 33
3.1.1 Characterization of fixed-dose combinations ................................................................ 34
3.1.2 Body of evidence ............................................................................................................ 34
P a g e | 10
3.1.3 Employed strategies in development ............................................................................ 35
3.1.4 Conclusions ..................................................................................................................... 36
3.2 Paper 2 - Demonstrating Contribution of Components of Fixed-Dose Drug Combinations
Through Longitudinal Exposure-Response Analysis ...................................................................... 37
3.2.1 Analysis strategy ............................................................................................................... 37
3.2.2 Exposure-response analysis .............................................................................................. 37
3.2.3 Clinical trial parameters .................................................................................................... 40
3.2.4 Discussion .......................................................................................................................... 40
3.2.5 Conclusion ......................................................................................................................... 40
3.3 Paper 3 - Pharmacodynamic modelling reveals synergistic interaction between docetaxel
and SCO-101 in a docetaxel-resistant triple negative breast cancer cell line ............................... 41
3.3.1 Monotherapy models ..................................................................................................... 41
3.3.2 Combination model ........................................................................................................ 42
3.3.3 Discussion ....................................................................................................................... 44
3.3.4 Conclusion ...................................................................................................................... 45
4 Conclusions and Perspectives for further research ..................................................................... 46
5 References .................................................................................................................................... 49
Papers ................................................................................................................................................. 59
P a g e | 11
Table of Figures Figure 1 – Schematic to guide the feasibility of fixed-dose combination development with two or more approved
compounds. FDC: Fixed-dose combination; NI: No Interaction; PD: Pharmacodynamic; PK: Pharmacokinetic; SI:
Significant Interaction. The figure is from [59]. ...................................................................................................... 20
Figure 2 – The origins of hyperglycemia in type 2 diabetes mellitus. Hyperglycemia arises from multiple sources of
dysregulated carbohydrate regulation. Each dysregulated function represents potential drug targets for
reestablishing the normal carbohydrate levels in the bloodstream. SGLT2, sodium/glucose co-transporter 2;
AMPK, AMP-activated protein kinase; DPP4, dipeptidyl peptidase 4; IκB, inhibitor of NF-κB; MAPK, mitogen-
activated protein kinase; NF-κB, nuclear factor-κB; RA, receptor agonist; ROS, reactive oxygen species; TLR4, Toll-
like receptor 4; TNF, tumor necrosis factor; TZDs, thiazolidinediones. The figure is from [64].............................. 23
Figure 3 – Primary mechanisms leading to drug resistance of the cancer cells. The primary mechanisms include inherent
cell heterogeneity, alterations in drug targets, drug efflux, and morphological changes in the form of epithelial-to-
mesenchymal transition. EMT, epithelial to mesenchymal transition. The figure is from [76]. ............................. 25
Figure 4 – Overview of indirect response model with inhibition of kin. kin, formation constant of endpoint (response
variable), kout, elimination rate constant of endpoint (response variable). ............................................................ 28
Figure 5 – Overview of two effect-based approaches: Response Additivity (left) and Bliss Independence (right). Drug A
effect: 30, Drug B effect: 20, Drug A+B effect: 48. .................................................................................................. 30
Figure 6 – Flowchart showing the identification of excluded and included European Public Assessment Reports (EPARs)
in paper 1. Distribution of the included EPARs in the anatomical therapeutic chemical (ATC) classification system
is shown to the left. The condition for “special exclusion condition” are outlined in Appendix S1 of paper 1 [14].
The figure and caption are from [14]. ..................................................................................................................... 33
Figure 7 – Flow-chart illustrating the process of the study. Exposure as trough concentrations and response as HbA1c
levels are sampled pairwise from the simulated data of each scenario. The samples are then analyzed using
exposure-response and longitudinal exposure-response models, and the false positive and false negative rates are
computed. Lastly, clinical trial parameters across all scenarios are analyzed. The figure and caption are from paper
2. ............................................................................................................................................................................. 37
Figure 8 – Pharmacodynamic response (HbA1c [%]) vs. compound A (left panel) and B through concentration (middle
panel) on a linear and logarithmic scale (embedded graph) as well as PD response vs. time (right panel) across four
scenarios. Each scenario includes; 60 subjects, 8 samples, and a trial duration of 100 days (A+B), 30 subjects, 4
samples, and a trial duration of 20 days (C+D). All scenarios use a multiple dosing scheme and a dose distribution
on either side of IC50. For scenario A+C, compound A is active, while for B+D, compound B is active. Each data
pair is represented by dots colored by dose level. The dashed line represents the loess line, and the solid line
reflects the underlying relationship between PD response and compound concentration. The figure and caption
are from paper 2. .................................................................................................................................................... 39
Figure 9 – MDA-MD-231 survival following monotherapy with docetaxel (A+B) or SCO-101 (C+D). The blue dots represent
the samples from the experiment, while the black line corresponds to average cell survival. The red dashed line
represents the fitted curve for docetaxel (Imax model) and SCO-101 (sigmoid Imax model). B and D contain the same
data as A and C, respectively, but on a logarithmic scale. The figure and caption are from paper 3 [127]............ 42
Figure 10 – Contour plot of the final general pharmacodynamic interaction model. The lines indicate four separate
response levels of 0.1 OD, 0.07 OD, 0.05 OD, and 0.03 OD across the dose combination space. 0.05 OD corresponds
to the 85% reduction in cell viability target. The four dots highlight dose pairs that result in meeting this target,
through lowest total dose combination (red), minimized exposure to both compounds (green), and with a
weighted penalty factor of 2 on lowest total dose combination (orange) and minimized exposure (yellow). The
figure and caption are from paper 3 [127]. ............................................................................................................. 43
P a g e | 12
1 Aim & Objectives
Combination drug therapy is becoming a central part of the treatment strategies for many diseases
where monotherapy so far has not given sufficient efficacy or where resistance to treatment
emerges [1–3]. Thus, more drug combination products can be expected to be developed in the
future. The standard practice for verification of efficacy for fixed-dose combinations is the factorial
design study in which the individual components are compared to the combination across all
investigated dose levels [4]. Given the trend towards personalized medicine, it is increasingly
important to discover alternatives to the factorial design study [5]. The factorial clinical trial is costly,
and it is challenging to perform this type of clinical trial when the number of investigated dose levels
increase [5]. To ensure lower developmental costs and easier access to the market, it is important
to avoid superfluous investigations by performing the correct and necessary clinical experiments
from the beginning. To that end, in silico methods provide tools that can assist in parts of the
development process and provided valuable information to guide the drug development, thereby
saving time and resources [6–9]. However, the tools and models for analyzing fixed-dose
combinations are only in their infancy.
Thus, the focus of this project and the overall aim is to evaluate, develop, and validate modeling
tools for the development of fixed-dose combination products.
1.1 Objectives
The specific objectives of the project were:
1. To assess the current practice of the use of modeling as a drug development tool
for the development of fixed-dose combinations in the European Union, based on
the information available in European Public Assessment Reports (EPARs)
published by the European Medicines Agency (EMA)
2. To develop new methods that may assist in the drug development process of fixed-
dose combinations, thus providing novel methodologies, which may be considered
by the relevant regulatory authorities when approving fixed-dose combinations
3. To validate combination models on preclinical data of a combination under
investigation and assessing the value of applying these models in guiding further
investigations of the combination
P a g e | 13
2 Introduction In the present thesis, boundaries were established for the objectives to facilitate focused
investigations. Overall, both combination therapies and fixed-dose combinations were considered
part of the scope for the project, depending on the objective that was considered. Specifically, when
concerning research questions that arose from legislation about drug development, only fixed-dose
combinations were considered as part of the scope. This was decided as the development of fixed-
dose combinations are addressed in a separate set of guidelines. Thus, for the first and second
objectives of the project, only fixed-dose combinations are evaluated. For the third objective, no
distinction was made between combination therapies and fixed-dose combinations. Both
combination therapies and fixed-dose combinations were included, since the mathematical
modeling of two pharmaceuticals in combination is, in general, applied in the same way, regardless
of being administered as two separate components or as a fixed-dose combination [10].
Several markets could be selected to assess the first objective of the project. The European market
in the form of submission to individual countries, the EMA, and the American market in the form of
submissions to the Food and Drug Administration (FDA) were all considered. The EMA was chosen
as the focal point of the discussion as the FDA has previously been the subject of several publications
on the topic [11–13]. Five previously published studies on the analysis of fixed-dose combination
approvals are summarized in Table 1. Furthermore, data from EMA is readily available to the public
through the EPARs. Despite not being the aim for this objective, the FDA was used as a comparison
for discussion.
Several disease areas could have been considered when analyzing fixed-dose combinations, the
most prominent being cardiovascular, metabolic, and infectious diseases [5]. An even wider scope
could have been considered when addressing combination therapy in its entirety. Cancer is a disease
area in which combination therapy is paramount to treatment success [14]. Here, greater efficacy
is in focus, but circumventing the emergence of resistance to treatment is equally or more essential
[14].
For the second objective, the regulatory guidelines set forth by the EMA is used as a starting point,
similarly to that of the first objective. Furthermore, a combination used for the treatment of
diabetes is chosen as it represents a major group of fixed-dose combinations under development.
In the third and final objective, a combination for the treatment of cancer is analyzed, as these
represent a complex and multifaceted combination in a field were combination treatment is of
major importance.
The topics, which have been briefly introduced above, are presented and discussed in greater detail
in the following sections.
P a g e | 14
Table 1 – Summary of five previously published studies on the analysis of fixed-dose combination approvals
Study Aim Main finding(s)
Fixed-Dose Combination
Drug Approvals, Patents,
and Market Exclusivities
Compared to Single Active
Ingredient Pharmaceuticals
[11]. (2015).
To assess the exclusivity life of
fixed-dose combinations and
the time between approval of
single drugs and the
corresponding fixed-dose
combination approved by the
FDA.
Fixed-dose combinations were approved
5.43 years after the single drugs. They
entered the market 2.33 years prior to
generic single drugs. Lastly, they added
9.70 years to the exclusivity life.
Fixed-dose combination
and single active ingredient
drugs: A comparative cost
analysis [12]. (2016).
To assess price differences
and pricing structures of fixed-
dose combinations and
corresponding single drugs
approved by the FDA.
The fixed-dose combinations were on
average 83.3 ± 23.4% of the cost of the
single drug, based on average wholesale
price. This difference in price was
correlated with the year of approval, the
number of generics available of
components in the combination, and the
therapeutic class.
An Analysis of the Fixed-
Dose Combinations
Authorized by the
European Union, 2009-2014
[15]. (2015).
To assess the effects of the
negative connotations
associated with fixed-dose
combinations following the
bans in the mid- to late 20th
century as well as
characterizing the reasoning
for the Authorization by the
EMA.
Stricter guidelines and regulations were
found for fixed-dose combinations
following the bans. Examples of
regulatory flexibility was seen when
given proper justification. The main
reason for authorization was increased
efficacy.
Analysis of Fixed-Dose
Combination Products
Approved by the US Food
and Drug Administration,
2010-2015: Implications for
Designing a Regulatory
Shortcut to New Drug
Application [13]. (2017).
To assess the development
process of fixed-dose
combinations approved by the
FDA and guide the
development of these
products in the US.
Approval of the fixed-dose combinations
was granted even if the full phases of
clinical development was not
completed. This highlighted a
development strategy where phases can
be exempted if proper justification is
provided.
Investigation of Approval
Trends and Benefits of New
Fixed-Dose Combination
Drugs in Japan [16]. (2019).
To assess the trends and
benefits of fixed-dose
combinations approved in
Japan using a questionnaire
survey
Cardiovascular agents were the largest
therapeutic group. Compliance of
bronchial asthma patients improved by
30.8% when taking fixed-dose
combinations. Prescribers reported a
decreased time and effort to prescribe
the fixed-dose combinations.
The study title, aim, and the main finding(s) of the study are presented for each of the five previously published studies on the analysis of fixed-dose combination approvals.
P a g e | 15
2.1 Fixed-dose combinations
Fixed-dose drug combinations are a subset of combination therapies containing two or more active
ingredients in a single dosage form. Combining two or more active ingredients in a single dosage
form enables distinct advantages and challenges for the development and use of fixed-dose
combinations. The following section outlines the current stance on the advantages, use, and
development of fixed-dose combinations as well as the rationale for combination therapies.
2.1.1 Combination therapy
Combination therapies involve the concurrent treatment of diseases with two or more
pharmacologically active compounds. The underlying principle for this approach is the ability to
target several biological pathways simultaneously [1, 17]. Advances in high-throughput genome
sequencing, loss- and gain-of-function methods, and several screening approaches have allowed
increased ability to describe these underlying biological pathways in a healthy and disease state [18,
19]. This description leads to a greater understanding of the disease and the possible biological
redundancies that are in place, as well as the potential for the development of resistance to
treatment [17]. Additionally, the understanding of diseases as the product of a perturbed network
of biological pathways has enabled the identification of molecular targets for combination therapies
[20].
There exist many purposes for targeting multiple pathways simultaneously. Pharmacokinetic drug
interactions are, in many cases, undesirable as they can alter the exposure of co-administered drugs
and thereby cause adverse events and/or treatment failure [21]. However, some combination
therapies are specifically developed to alter the pharmacokinetics of a compound. The combination
of cobicistat and darunavir for the treatment of Human immunodeficiency virus (HIV) makes use of
this principle. Cobicistat inhibits CYP3A4 and CYP2D6, which are the main metabolizers of darunavir,
thereby increasing the exposure of darunavir [22].
Other combination therapies make explicit use of the components not interacting directly. In the
treatment of the cancer subtype, diffuse large B-cell lymphoma, the use of a 5-component chemo-
immunotherapy agent termed R-CHOP is the current standard of care [23]. The advantage of this
combination arises from targeting separate pathways, thereby limiting the cross-resistance that can
develop to the treatment [24].
Another approach to take advantage of targeting multiple pathways at the same time is by affecting
a specific biomarker from multiple angles. Targeting multiple pathways is often the case for
diabetes, where the regulation of blood glucose is central to controlling the disease. In combinations
for treating diabetes, pathways that affect the distribution and excretion of glucose are often
targeted. In the case of the linagliptin and empagliflozin combination, linagliptin causes lower blood
glucose through increased blood concentration of incretins and empagliflozin inhibits reuptake of
glucose in the kidneys, thereby increasing the excretion of glucose. By administering the compounds
together, a greater efficacy is achieved through a synergistic pharmacodynamic effect [25].
Thus, the development of combination therapies have been subject to increased focus in recent
years [2, 3] and has become an essential part of the successful treatment of several diseases [1].
P a g e | 16
Examples of the importance of combination therapy can be found as far back as the 1950s, where
the combination of streptomycin and isoniazid led to greater cure rates of tuberculosis [26]. A more
recent example is the extensive use of combination therapies for the treatment of HIV. The
introduction of the Highly Active Antiviral Retrovirus Therapy (HAART) caused HIV to be considered
a chronic illness rather than a death sentence. However, HAART also causes notable adverse events,
and therefore, HAART highlights one of the limitations of combination therapy, namely toxicity [27].
Several other disease areas make use of combination therapies with notable areas, including cancer,
cardiovascular disease, and metabolic diseases. An interesting perspective from cardiovascular
diseases is the association with metabolic diseases. Due to this association, the combinations are
often aimed at addressing the underlying issues for both the cardiovascular and metabolic disease,
such as high blood cholesterol [28]. Cancer and metabolic diseases will be addressed in further detail
in section 2.2.
Fixed-dose combinations utilize the same principles that make combination therapy essential. The
class of compounds can be considered a subset of the overall combination therapy group, with fixed-
dose combinations being different by being formulated as a single dosage form. This aspect provides
both advantages and disadvantages, which are largely dependent on the disease area.
2.1.2 Fixed-dose combination background
Fixed-dose combinations exist in many different forms. They are present as products ranging from
household ware to advanced medicines. One of the most common fixed-dose combinations are
probably the multivitamin preparations. These contain upward of 20 active ingredients and
represent a rather extreme example of fixed-dose combinations. WHO’s list of essential medicines
includes approximately 40 drug combinations of which most are available as fixed-dose
combinations [29]. However, the WHO states that “not all of the fixed-dose combinations in the
WHO treatment guidelines exist, and encourages their development and rigorous testing.” [29].
Despite being common, the reputation of fixed-dose combinations suffered a setback in the 1950s
[30]. This was brought about by combining diuretics with potassium chloride. The underlying reason
for making this combination is that a side effect of diuretics is the removal of potassium in addition
to the intended sodium removal. Thus, potassium tablets were often given in conjunction with the
diuretic. However, when the fixed-dose combination of the two was administered, it resulted in
several cases of punctured stomach lining, which required surgery [30]. Thus, fixed-dose
combinations were subsequently associated with negative connotations [30].
More recently, a similar issue related to the reputation of fixed-dose combinations has occurred in
India. Here, the development of fixed-dose combinations had been very popular, with a total of
1306 approvals in the period from 1961 till 2019 across all therapeutic areas [31]. The rationale
behind a large amount of these approvals have been insufficient, as many combinations were
developed either solely for marketing interests or were developed with neither theoretical
justification nor evidence [32]. Over the last decade, this had led to 294 licenses being withdrawn,
the use of 344 fixed-dose combinations being prohibited, and 328 bans being issued by the central
government [31]. The issues in India underlines that while combinations can be a powerful tool;
P a g e | 17
ensuring the validity of the underlying rationale for combining pharmaceutical compounds is of
utmost importance.
In other areas of the world with greater cognizance of the regulatory framework, such as the United
States of America with the FDA and the European Union with the EMA, fixed-dose combinations
have seen a rise in approvals [5, 11, 13, 16]. The reemergence of fixed-dose combinations and the
revitalized focus on developing them could stem from an unmet medical need, as monotherapies
provide insufficient efficacy in several disease areas. However, filling this unmet medical need is not
exclusive to fixed-dose combinations, as all combination therapies can fill this gap. The following
section addresses advantages, which are tied to fixed-dose combinations.
2.1.3 Advantages of fixed-dose combination
There are several areas where fixed-dose combinations show advantages over the free-equivalent
combinations (FEC). FEC constitutes the same active ingredients as the fixed-dose combination, but
in two or more dosage forms. One of the main clinical advantages is the increased adherence to
treatment due to a lower medication burden, which has been demonstrated in numerous studies
and across several disease areas [33, 34]. In an example from type 2 diabetes mellitus (T2DM), an
analysis of seven studies was conducted, where the fixed-dose combinations were compared to the
FEC. Here the fixed-dose combinations were associated with a 13% greater adherence [35]. Another
example within T2DM assessed the HbA1c levels in 6000 European patients across multiple
nationalities, taking either fixed-dose combination or the FEC [36]. Here the advantage of taking the
fixed-dose combination manifested through a 0.25% lower level of HbA1c, thus providing better
glycemic control than the FEC counterpart [36]. Another interesting point in the study was a
comparison between patients classified as either compliant or non-compliant. Here, the compliant
patients were found to be five times more likely to be taking the fixed-dose combination than the
non-compliant patients. Similar results for increased adherence has been seen within the treatment
of hypertension [34, 37] and cardiovascular disease [38].
So far, only the clinical benefits of fixed-dose combinations have been discussed. Another, less
obvious advantage, which creates focus on fixed-dose combination development, is the extension
of patents.
When inventing novel drugs, a patent ensures the protection and market exclusivity for the
developing company for 20 years [39]. A substantial amount of this period can be spent in the
developing phase, during testing and clinical trials. Following development, companies typically
have between 7-12 years to make a profit on the investment, before generic drug companies enter
the market [40]. Here, fixed-dose combinations can provide an avenue to extend the commercial
lifespan of the compound by combining the newly patent expired compound with another non-
patented or patented compound. However, this strategy is dependent on either the newly
combined product being superior to existing treatment, thereby enabling the option for obtaining
a patent for the combination, or convincing prescribers to switch to the combination, which would
not currently have competition from generic medicines.
P a g e | 18
Several strategies exist for minimizing the commercial risk of developing fixed-dose combinations
and providing a superior product. Combining patent expired product with the current standard of
care treatment is a common approach. For instance, this is the basis for many fixed-dose
combinations within diabetes care. Here metformin is combined with a range of different products,
resulting in a convenience product consisting of the patent expired best-in-class drug and the
standard treatment [5, 41]. Another strategy is the collaboration of competing companies to create
new fixed-dose combinations using their patent expired stand-alone products. This approach has
been shown to be particularly successful [42]. Examples of this include Atripla®, a three-component
fixed-dose combination for the treatment of HIV. Competing companies Gilead and Bristol-Myers
Squibb developed Atripla®, which in the first six months after launch, had achieved sales for over
$200 million [42]. An example within hyperlipidemia is the joint development of Vytorin® by Merck
and Schering–Plough, which combined ezetimibe and simvastatin to lower blood cholesterol
concentrations [42].
Allowing the purchase of both the fixed-dose combination and the separate components seems to
be an important consideration. This was showcased by the public backlash Pfizer faced in 2006 when
they planned to discontinue torcetrapib and only sell it as part of their new fixed-dose combination
Lipitor® [43].
Pursuing the option of convincing prescribers to switch to the new combined product will be largely
dependent on cost. There are several cost-related advantages of fixed-dose combinations. Firstly,
the cost of the fixed-dose combination is commonly comparable to or lower than the total cost of
the components [33]. Secondly, the co-pay by the patients may be reduced, as only one co-pay is
required for the fixed-dose combination in comparison to the several that may be required for the
components separately [33]. An important consideration when considering cost is the different
national reimbursement systems that exist within the EU [44]. It is usually local committees, which
determine if a fixed-dose combination is eligible for reimbursement [44]. Thus, patients are not
directly in control of whether they receive the FEC or fixed-dose combination as they are influenced
by the decisions of the local health care systems. The various reimbursement systems are very
complex [44], therefore, further discussion of the topic is considered outside the scope of this thesis.
It has been demonstrated within antihypertensive medication that patients on a fixed-dose
combination had a lower prescription and total medical cost compared to the patients on the same
medication but in FEC [45]. A broader analysis across multiple therapeutic areas found similar
results, that is that the average wholesale price of fixed-dose combinations are on average lower
than the total cost of the FEC [12]. Interestingly, this analysis highlighted that the difference in cost
between the fixed-dose combination and the FEC is highly dependent on the therapeutic area.
Furthermore, the study analyzed the price point selection of the fixed-dose combination and
determined that it was most often set to the price of the costliest single agent in the combination.
Selecting this price point was done in order to shift the demand from generics towards the fixed-
dose combination [12].
P a g e | 19
It has been demonstrated, in some cases, that the immediate out-of-pocket costs of fixed-dose
combinations may be higher than for their FEC, but that the total long term or intangible costs were
lower [46, 47]. Furthermore, it has been shown that a fixed-dose combination can lead to fewer
clinic visits and laboratory tests, which further reduces the overall cost of treatment [47]. However,
the opposite pattern has also been observed in some cases. Here the total cost of fixed-dose
combinations is initially lower, but once the generics enter the market, the FEC becomes the
cheaper option [48]. An important consideration here is that an increase in cost has been shown to
associate with lower adherence [49]. Thus, identifying the cheapest option for patients is important
for ensuring adequate treatment.
Overall, analyzing the marketing strategy for fixed-dose combinations with regards to costs is
multifaceted. Providing a better standard of care through combining the best in class products with
the standard care in a fixed-dose combination and making it available at the same or lower cost than
the FEC is a major selling point of fixed-dose combinations. However, it can be argued that the
strategy of extending patents through the development of fixed-dose combination can lead to a
greater degree of questionable marketing strategies. In particular, this could be the case if ensuring
the extension of the product lifespan is the sole motivation behind development.
2.1.4 Development of Fixed-dose combinations
The International Council for Harmonisation of Technical Requirements for Pharmaceuticals for
Human Use (ICH) is an association that works to harmonize regulatory policy across its 16 members
and 32 observers [50]. Consequently, guidelines developed by agencies such as the EMA and FDA
are highly influenced by the ICH guidelines. The development of fixed-dose combinations is subject
to a separate set of guidelines from the conventional drug development guidelines [51, 52].
Currently, there is no ICH guideline specific to combination drugs, instead, the ICH monotherapy
guidelines serve as a framework for the development of fixed-dose combinations [53].
Each of the major agencies, which govern drug development, has its own distinct set of guidelines
on the development of fixed-dose combinations, which has been the topic of several publications
[41, 54]. In the European Union, the EMA has compiled the guideline “Guideline on clinical
development of fixed combination medicinal products”, which came into effect in 2017 [51]. In the
United States of America, the FDA provides two guidance documents relating to fixed-dose
combinations and combination therapy. For fixed-dose combinations, the document “Fixed Dose
Combinations, Co-Packaged Drug Products, and Single-Entity Versions of Previously Approved
Antiretrovirals for the Treatment of HIV” from 2006 and for combinations therapy the document
“Codevelopment of Two or More New Investigational Drugs for Use in Combination” from 2013 [52,
55]. Lastly, the World Health Organization (WHO) has written a guideline for fixed-dose
combinations “Guidelines for registration of fixed-dose combination medicinal products” from 2005
[56].
The overarching theme for all the guidelines on fixed-dose combination is summarized well in the
EMA’s three overall requirements [51]:
1. Justification and rationale for the combination
P a g e | 20
2. Demonstrating the contribution of all active substances to the desired therapeutic effect
3. The evidence presented is relevant to the fixed combination medicinal product (important if the
evidence is based on the administration of separate active substances in combination)
Another important consideration made by these agencies is based on the approval status of the
components of the fixed-dose combination. Here four scenarios outlined in the WHO guideline
provide a summary that matches the consideration made in all the guidelines [56]. The four
scenarios of fixed-dose combinations that can be considered for approval are summarized here:
1. Generic fixed-dose combination of an already existing fixed-dose combination
2. Fixed-dose combination in the same dosage as two separate compounds that are administered
as part of an existing treatment regimen
3. Fixed-dose combination of two compounds not previously combined or combined in new
dosage regimen
4. Fixed-dose combination containing one or more new chemical entities
Figure 1 – Schematic to guide the feasibility of fixed-dose combination development with two or more approved compounds. FDC: Fixed-dose combination; NI: No Interaction; PD: Pharmacodynamic; PK: Pharmacokinetic; SI: Significant Interaction. The figure is from [57].
P a g e | 21
Justification for the combination can range from an improvement in benefit/risk to utilizing drug-
drug interactions that, for instance, circumvent resistance development or cause non-efficacious
compounds to become effective [51]. Examples of this are mentioned in section 2.1.1. Importantly,
simplification of the treatment is not sufficient to constitute a rationale for the combination [51].
Based on each of the four scenarios, and the justification behind combining the components, the
evidence base supporting the combination can be required to be appreciably larger or smaller.
Specifically, it has been shown that the evidence base, measured as the number of patients, arms,
and clinical trials, increases when one or more new molecular entities are included in the fixed-dose
combination [5].
The simple scenario involves combining two already approved compounds in a fixed-dose
combination. A schematic of a decision-making process for developing this type of fixed-dose
combination is shown in Figure 1.
As seen in the schematic, for this type of fixed-dose combination, the most important aspect
becomes to determine if drug-drug interactions occur in the pharmacokinetics (PK) or
pharmacodynamics (PD) of the compounds. The purpose of investigating drug-drug interactions in
PK is to determine if the LADME profiles of the compounds are altered in each other’s presence [21,
58]. For fixed-dose combinations, the general expectation is that the presence of both compounds
do not affect the PK profiles [57]. However, in some distinct cases, the purpose of the combination
is to affect the PK profile of one compound, such as for cobicistat and darunavir discussed in section
2.1.1.
PD interactions are much more complicated to analyze. PD interactions can broadly be categorized
into three categories.
The first describes fixed-dose combination with no PD interactions, which would
theoretically provide at higher efficacy than either of the components alone
The second includes fixed-dose combinations with a negative interaction, which would result
in terminating the development process.
The third group constitutes fixed-dose combinations with a positive interaction, which is the
most promising group of drug-drug combinations. This group is arguably the most innovative
combinations, which utilize drug-drug interaction to achieve higher efficacy or safety.
A more complex scenario than combining two previously approved compounds is to develop one or
more new molecular entities as part of a fixed-dose combination. Of the total two-component fixed-
dose combination approved from 2010-2016 by EMA, 36% included one new molecular entity and
5.5% included two new molecular entities in the combination [5]. Ideally, the focus when developing
this type of fixed-dose combination is to establish the safety and efficacy of the components
separately before analyzing the combination. Thereby, the development follows a similar pattern
to using approved compounds after the safety and efficacy are established for the monotherapies.
There may however be cases where this is not feasible. This is the case when the components alone
cause rapid resistance development in certain diseases such as HIV and microbial diseases [51, 55].
P a g e | 22
Demonstrating the contribution of each of the components to the overall effect is a central part of
adhering to the requirements for fixed-dose combination development. This is the second overall
requirement outlined earlier from the EMA and referred to in article §300.50 “Fixed-combination
prescription drugs for humans” from the FDA, which is often termed “the combination rule” [51,
59]. The standard approach to achieve this is to perform a factorial clinical trial in which multiple
doses of each component is compared to the combination, and the effect of A+B is demonstrated
to be greater than that of either A or B [4]. This approach is obviously not feasible for the combined
new molecular entities causing resistance when administered separately, and therefore, the
approach is here to compare the combination to the standard of care [51, 55].
Other considerations for the development of fixed-dose combination include bioequivalence
studies and the dosing intervals of the components. As part of the development package for fixed-
dose combinations, it is expected to demonstrate the bioequivalence of the fixed-dose combination
to each of the components [51, 52, 57]. For fixed-dose combinations composed of previously
approved drugs, the bioequivalence studies can make up a substantial part of the development
program, as other parts of the development program, such as dose-finding, can be excluded [5].
Achieving bioequivalence for a fixed-dose combination presents several challenges depending on
the differences in dosage forms, formulations, and LADME profiles of the fixed-dose combination
and the individual mono-components [60]. Depending on these differences it can be considered
significantly more challenging to demonstrate bioequivalence between a fixed-dose combination
and the individual mono-components than between two formulations of the same active ingredient
[60]. While bioequivalence is important for fixed-dose combination development overall, the topic
is not part of any studies conducted in the present thesis and is therefore not discussed in further
detail. Dosing intervals present an issue when combining compounds with very different half-lives,
as it often results in different frequency of administration. Thus, a modification of the regimen will
be necessary for development to continue. Similarly, opposite administration conditions, such as
fasting/fed condition, can cause the development to be unfeasible.
2.1.5 European Public Assessment Reports
EPARs are a public compilation of documents published by EMA, which pertain to the evaluation of
drug authorizations through the centralized procedures at EMA [61]. The documents include
information on “clinical aspects”, “clinical efficacy”, and “clinical safety”, as well as “Product
information”, and “Authorized presentations”. While not specifically relating to fixed-dose
combinations, the documents were used as the basis for the first paper discussed in the summary
of results (section 3.1).
P a g e | 23
2.2 Combination therapy in diabetes & breast cancer
The present thesis is not limited to specific disease areas; however, during the experimental phase,
two disease areas became the focus of the second and third papers. In this section, the two disease
areas, diabetes and breast cancer, will be presented briefly in the context of combination therapy
with a focus on the areas important to the research conducted in this thesis.
2.2.1 Type 2 diabetes mellitus
T2DM is a metabolic disease characterized by failure to regulate carbohydrate levels in the
bloodstream [62]. T2DM constitutes more than 90% of all diabetes cases and is, therefore, the most
common subtype [62]. The dysregulation of the carbohydrates arises from several underlying
mechanisms, Figure 2. The understanding of these mechanisms is rapidly evolving. Currently, the
understanding is that the main cause is a combination of insulin resistance in skeletal muscle, liver,
and adipose tissue, coupled with increasingly impaired insulin secretion from the pancreas [63]. This
leads to a condition known as prediabetes and predisposes for the development of T2DM, which is
hallmarked by hyperglycemia [63, 64].
Figure 2 – The origins of hyperglycemia in type 2 diabetes mellitus. Hyperglycemia arises from multiple sources of dysregulated carbohydrate regulation. Each dysregulated function represents potential drug targets for reestablishing the normal carbohydrate levels in the bloodstream. SGLT2, sodium/glucose co-transporter 2; AMPK, AMP-activated protein kinase; DPP4, dipeptidyl peptidase 4; IκB, inhibitor of NF-κB; MAPK, mitogen-activated protein kinase; NF-κB, nuclear factor-κB; RA, receptor agonist; ROS, reactive oxygen species; TLR4, Toll-like receptor 4; TNF, tumor necrosis factor; TZDs, thiazolidinediones. The figure is from [62].
As the illustration indicates, the pathogenesis of T2DM is complex and arises from multiple
metabolic defects [62, 63]. Hyperglycemia arising from these metabolic defects is known to be a
large factor in the risk of complications from diabetes [65]. Therefore, reaching normal blood
glucose levels, often measured as the biomarker HbA1c (signifying glycosylated haemoglobin in
blood cells), through blood glucose-lowering treatment is incredibly important [66]. Achieving this
P a g e | 24
glycemic target is the main goal of T2DM treatment through GLP1 analogs, DPP4 inhibitors, SGLT2
inhibitors, thiazolidinediones, and metformin, Figure 2. However, there are issues with reaching the
general goal of 7% HbA1c in many cases, as illustrated by a study in the U.S. from 2007-2010, where
only 52.5% of individuals reached the target [67]. Hence, initiating therapy with only a single drug
has been suggested not to be sufficient to ensure proper glycemic control [63]. Thus, targeting
multiple parts of the pathogenic network is of interest, as it has been demonstrated to achieve
increased glycemic control [66].
A counterpoint to combination treatment T2DM is the heterogeneity of the disease that arises from
the complex pathogenic network. For instance, some patients (5-10%) cannot tolerate metformin
treatment [68] and would, therefore, need alternative combination products to achieve their
glycemic targets.
2.2.2 Triple-negative breast cancer
Triple-negative breast cancer (TNBC) constitutes 15-20% of all breast cancers and is associated with
onset at an early age, aggressive clinical course, and a very poor prognosis in comparison to the
hormone receptor and HER2 positive breast cancers [69]. TNBC has a unique pattern of metastasis
to other organs compared to the two other subtypes [70]. The pattern is characterized by a larger
degree of brain and lung involvement and a lower degree of bone lesions [70]. Subtyping cancers
based on protein expression and phenotype is an ongoing process, which for TNBC has led to
additional subtyping into categories based on molecular characteristics [71]. Further subtyping is,
however, outside the scope of TNBC discussion for this thesis.
In the early stages of TNBC, the cancer cells are generally chemotherapy-sensitive; however, there
is no defined optimal treatment strategy [72]. The most common initial step for more advanced or
inoperable TNBC is neoadjuvant therapy, which includes anthracyclines, taxanes (such as
docetaxel), and cyclophosphamide [72, 73]. Given the severity of the disease, there is a constant
effort to identify and utilize new therapies for TNBC patients [69].
As seen from the neoadjuvant therapy, combinations of two or more compounds that target specific
pathways in the cancer cell is a cornerstone in cancer therapy [74].
Drug resistance is a multifaceted phenomenon that spans inherent cell heterogeneity, alterations
in drug targets, drug efflux, and morphological changes in the form of epithelial-to-mesenchymal
transition, Figure 3. Due to the wide range of resistance mechanisms, combination therapy is
considered as the best treatment option for all cancer. This is due to a lower risk of drug resistance,
and because the combined therapies will generally be more effective than the monotherapies [74].
A general goal of combination therapy in cancer is to utilize drugs with different primary
mechanisms, leading to low shared cross-resistance between the compounds [24].
P a g e | 25
Figure 3 – Primary mechanisms leading to drug resistance of the cancer cells. The primary mechanisms include inherent cell heterogeneity, alterations in drug targets, drug efflux, and morphological changes in the form of epithelial-to-mesenchymal transition. EMT, epithelial to mesenchymal transition. The figure is from [74].
Precision medicine further adds to the complexity of TNBC treatment. Precision medicine aims to
provide a tailored treatment to patients through the identification of not only general genetic
variants but also specific pathway and network alteration, which can govern highly specific
treatments of the individual cancer patient [75].
Based on the complexity and attempt of individualization of cancer treatment, it is no surprise that
0% of approved fixed-dose combinations are within the field of cancer [5, 76]. Fixing the dosage or
ratio between components is not feasible for combinations with the need for such a high degree of
individualization. However, for the purposes of analyzing and modeling drug-drug combinations, it
presents an interesting topic.
P a g e | 26
2.3 Pharmacokinetic-Pharmacodynamic Modelling & Simulation
Pharmacokinetic-Pharmacodynamic (PK-PD) modeling and simulation were a part of the second and
third paper in the present thesis. While standard model building in the form of sequential structural,
stochastic, and covariate modeling was not performed, the same principles were applied for
simulation and are therefore presented. Finally, analysis and modeling of data was performed using
exposure-response modeling and combination modeling, which is discussed in the following
sections.
2.3.1 Background
In essence, models provide a simplified and interpretable representation of systems that are of
interest to the modeler [77]. In PK-PD modeling, the system that the modeler attempts to describe
is the timecourse of drug concentration following administration (PK) as well as the link between
that concentration and the effect of the drug (PD) [77].
The value of modeling is substantial in all phases of drug development. In the preclinical phase, the
focus is on determining potential drug candidates. Here modeling provides a way to characterize
the potency, bioavailability, clearance, toxicity, and drug interactions [78]. In addition, it can be used
to guide optimal sampling and dose ranges for further investigation [78]. Establishing a model early
in development is essential for successfully utilizing modeling to guide drug development. The
importance of establishing the model early on is due to the ability to constantly update the model
through an iterative process as data become available [79].
Throughout the clinical development, spanning phase I to III, the scope of modeling expands. In the
early phases, there is a focus on accurately describing the PK relationships and estimating the
probability of success given assumptions and trial designs [78]. In the later phases, confirmation of
a valid PK-PD model and establishing the exposure-response relationship is in focus [78].
Importantly, these are all just examples of the goals that can be achieved using modeling but is not
exhaustive.
Within the regulatory agencies EMA and FDA, modeling and simulation are considered an important
tool to support rational decision making throughout the development process. The EMA has
released guidelines for physiologically based pharmacokinetic modeling and simulation as well as
appointed a working party, which focuses on outlining the standards for employing modeling and
simulation in drug development [6, 7]. Similarly, the FDA has published guidelines on
pharmacokinetic modeling (currently in draft version in 2019) [8]. Additionally, the FDA has
presented a talk on the topic termed “Advancing regulatory science with modeling and simulation
at FDA”, in which they recognize how modeling and simulation can benefit public health [9].
2.3.2 Population PK-PD modeling
Population PK-PD models consist of three overall components: structural models, stochastic models,
and covariate models.
The purpose of the structural model is to describe the timecourse of the PK or PK-PD relationship.
This description is achieved either through algebraic or differential equations [77]. In the case of PK
P a g e | 27
models, the structural model is often described by a compartmental model. A compartmental model
typically consists of one to three compartments, with additional absorption compartments for
extravascular administration. An example of the differential equation approach of a one-
compartment model with first-order absorption and elimination is shown in EQ1.
The differential equation describes the rate of change in drug amount for the absorption
“compartment” (1) and the central compartment (2). The equation is solved in very small time
increments in order to minimize the computational errors. The estimated population parameters,
here absorption rate constant kA, clearance (CL), and volume of distribution (V), from the structural
model, are often termed thetas (θ). Through this approach and developments within algorithms,
the difference between the algebraic solution and the differential equation is negligible but can
come with a time cost depending on the complexity of the model [77].
The PK is linked to the PD in a specified relationship, and through this relationship, the timecourse
of the drug response is described [80]. This relationship can either be direct or delayed depending
on the specification. For the direct PK-PD relationships, the plasma concentration is used directly in
the description of the drug response, with relationships often being linear, log-linear, or non-linear
(e.g., Emax relationship and sigmoidal Emax relationship), EQ2 [80]. The parameters in the models are:
α, the intercept with the y-axis; β, the slope; 𝐶𝑑𝑟𝑢𝑔, the drug concentration; 𝐸𝑚𝑎𝑥, the maximal effect;
𝐸𝐶50, the half-maximal concentration, and 𝐻𝑑𝑟𝑢𝑔, the Hill coefficient.
𝐸𝑑𝑟𝑢𝑔 = β ∙ 𝐶𝑑𝑟𝑢𝑔 + α; 𝐸𝑑𝑟𝑢𝑔 = β ∙ log (𝐶𝑑𝑟𝑢𝑔) + α;
𝐸𝑑𝑟𝑢𝑔 =𝐸𝑚𝑎𝑥,𝑑𝑟𝑢𝑔 ∙ 𝐶𝑑𝑟𝑢𝑔
𝐸𝐶50,𝑑𝑟𝑢𝑔 + 𝐶𝑑𝑟𝑢𝑔 ; 𝐸𝑑𝑟𝑢𝑔 =
𝐸𝑚𝑎𝑥,𝑑𝑟𝑢𝑔 ∙ 𝐶𝑑𝑟𝑢𝑔
𝐻𝑑𝑟𝑢𝑔
𝐸𝐶50,𝑑𝑟𝑢𝑔
𝐻𝑑𝑟𝑢𝑔 + 𝐶𝑑𝑟𝑢𝑔
𝐻𝑑𝑟𝑢𝑔
(EQ2)
For the delayed PK-PD relationships, the effect compartment model and the indirect response
model (turnover model) are well established [80, 81].
The effect compartment model is used when the site of drug action is different from the site of
measured drug concentration [80, 81]. For instance, this can be the case if the site of drug action is
in the brain. Thus, a delay in effect occurs, which is described by the first-order rate constant ke0
between the observed concentration and a hypothetical effect compartment [80, 81].
The indirect response attempts to capture the delay in effect that arises from targeting upstream
components of a signaling cascade. In this model, the effect is asserted through stimulation or
inhibition of the formation constant kin or the elimination rate constant kout of the response
(endpoint) [80, 81]. An example where kin is inhibited is shown in Figure 4. This model is used in the
simulations for the second paper of this thesis.
𝑑𝐴1
𝑑𝑡= −𝑘𝐴 ∙ 𝐴1
𝑑𝐴2
𝑑𝑡= 𝑘𝐴 ∙ 𝐴1 −
𝐶𝐿
𝑉∙ 𝐴2
(EQ1)
P a g e | 28
Figure 4 – Overview of indirect response model with inhibition of kin. kin, formation constant of endpoint (response variable), kout, elimination rate constant of endpoint (response variable).
The stochastic model attempts to capture the variability in the estimated parameters by adding
random effects to the model. Random effects can span numerous factors, such as inter-individual
variability, inter-study variability, and inter-occasion variability. The random effects are termed etas
(η), which are usually assumed to be distributed normally or log-normally around zero with a
variance (ω).
For simplicity, the following example includes a theoretical parameter 𝑃, which varies log-normally
between individuals, resulting in the individual parameter 𝑃𝑖, EQ3.
Note that while the value of 𝜂𝑃𝑖 is normally distributed, applying it exponentially to 𝑃 produces the
log-normal distribution.
While the random effects represent the random unexplained variability, the covariate model
attempts to explain the variability. Explaining the variability can be achieved through established
relationships, such as that between body weight and volume of distribution or more specific
considerations such as genotypes [77]. Covariate modeling is not a focus in this thesis; therefore,
the topic is not introduced further.
Determining the optimal model fit is based on model evaluation. Model evaluation can be based on
numerous different numerical and graphical approaches. Some of the common numerical
considerations are the Akaike information criterion (AIC), the Bayesian information criterion (BIC),
and objective function value (OFV) [82, 83]. These values are used for model comparisons in various
scenarios and represent how well the model fits the data. Parameter precision, in the form of
standard errors, confidence intervals, bias, and shrinkage, are important for assessing model
stability and robustness [82, 83]. For instance, high parameter imprecision can be a sign of an
overparameterized model [82]. A general consideration for high imprecision is >30% SE for fixed
effect and >50% SE for random effects [82]. Graphical diagnostics of the model are employed to
identify model misspecification [83]. Examples of graphical diagnostics include conditional weighted
residuals (CWRES) against independent variables (e.g., time, concentrations), observed versus
population/individual predictions, and simulation-based visual predictive checks (VPC) [82, 83].
Simulation from a PK-PD model is an important tool used for inference and model evaluation.
Simulating unobserved data within the bounds of the data for the model can be done with some
confidence, whereas simulating data outside the bounds requires confidence in the model used for
simulation [77]. Simulating from a model that includes a stochastic model to allow for unexplained
variability in the data requires the use of random number generators. An important consideration
𝑃𝑖 = 𝑃 ∙ 𝑒𝜂𝑃𝑖 , 𝜂𝑃𝑖 ~ 𝑁(0, 𝜔𝑃2) (EQ3)
P a g e | 29
here is that when sampling from random distributions, the simulations need to be repeated [77]. In
order to summarize the results as confidence intervals, 1000 simulations are usually recommended
[77].
2.3.3 Exposure-response analysis
Exposure-response analysis attempts to describe a relationship between an exposure metric and a
measured response variable. Several exposure metrics can be considered for this purpose, once the
steady-state of the PK is reached. Common metrics include the area under the concentration-time
curve (AUC), maximal drug concentration (Cmax), or the trough concentration, depending on the
specific analysis [84]. The response variable is often a clinical endpoint, which is coupled to the
exposure metric through linear or non-linear relationships presented in EQ2. Additionally, the
exposure-response analysis can study either a single time point or be a timecourse analysis [84]. The
single time point is often sufficient and a more common approach. However, timecourse analysis
has advantages when inter-occasion variability is high, when the endpoint changes over time at
steady-state PK, or if the data includes a high degree of dropouts [84].
Exposure-response analysis has an advantage compared to pairwise comparisons due to the
intrinsic variability in the PK that gives rise to a series of exposures for a given dose level and sample
size [85]. This variability in exposures results in greater power to identify the exposure-response
relationship. However, the variability in the exposures also constitutes a limitation of the analysis,
as there is a lack of randomization of the exposures, whereas the dose used in the pairwise
comparison is a controlled and randomized variable [85]. Another limitation is the susceptibility to
undetected confounders in the model [84]. Confounders in exposure-response analysis is the
implicit topic of the second paper in this thesis. Essentially the paper is based on the finding by Zhu
& Wang, which demonstrates that exposure-response analyses of fixed-dose combinations lead to
inflated false-positive rates [86]. Despite these limitations, exposure-response analysis is becoming
a tool with an increasing role in regulatory decisions [87–90].
2.3.4 Combination modeling
Combination modeling is performed to demonstrate that the effect of the combination is superior
to the effect of the components [20]. This is typically achieved through characterizing the
combination as either synergistic or antagonistic, corresponding to either higher or lower effect,
respectively, than the expected additive effect. Defining the additivity criterion can be the most
complex part of classifying a combination as synergistic or antagonistic, as opposing results can be
obtained depending on the criterion selected [20]. An example of these opposing results is
illustrated in Figure 5, where a synergistic combination within Bliss Independence can be considered
antagonistic in Response Additivity. This is evident from the combination being below the additivity
criterion (dashed line) for Response Additivity and above it for Bliss Independence, Figure 5.
P a g e | 30
Figure 5 – Overview of two effect-based approaches: Response Additivity (left) and Bliss Independence (right). Drug A effect: 30, Drug B effect: 20, Drug A+B effect: 48.
Response Additivity, Bliss Independence, and Loewe Additivity are three common effect-based
approaches for defining the additivity criterion [20, 91–93]. Loewe additivity and Bliss Independence
is based on opposing mechanistic assumptions surrounding the compounds in the combination [94].
Loewe Additivity assumes that both compounds share the mechanism of action, thereby making the
assumption that one compound can be substituted with the other. For Bliss Independence, the
assumption is that the compounds have different mechanisms of action, leading to the addition of
the component effects. Response Additivty rests on the principle that the effects of the combination
are simply additive and that the dose-effect relationship is linear [20]. While Loewe Additivity is a
common approach in combination modeling, it is not utilized in the present thesis, as the
investigated combination components do not share the mechanism of action. Loewe Additivity is
therefore not considered further. The mathematical equations for a combination of drug A and B
using Response Additivity and Bliss Independence is shown in EQ4 and EQ5, respectively.
𝐸𝑎𝑑𝑑𝑖𝑡𝑖𝑣𝑒,𝑅𝐴 = 𝐸𝐴 + 𝐸𝐵 (EQ4)
𝐸𝑎𝑑𝑑𝑖𝑡𝑖𝑣𝑒,𝐵𝐼 = 𝐸𝐴 + 𝐸𝐵 − 𝐸𝐴 ∙ 𝐸𝐵 (EQ5)
As given by EQ4, the additivity criterion for Response Additivity is the addition of the effect terms
for compound A and B, while any observed effect above or below that level corresponds to synergy
or antagonism, respectively [20]. For Bliss Independence, the additivity criterion is the product of
the effect term for compound A and B, subtracted from the sum of the two. Furthermore, the effect
terms are based on probabilities and are therefore constrained between 0 and 1. Thus, as the effect
of the components increase, the 𝐸𝑎𝑑𝑑𝑖𝑡𝑖𝑣𝑒,𝐵𝐼 in EQ5 will approach a combined effect of 1 [20, 92].
While underlying mechanistic assumptions are made for Bliss Independence, it is important to note
that there is no mechanistic basis for the model itself and therefore, it is solely an effect-based
approach.
P a g e | 31
Preclinical modeling of cancer chemotherapeutics has previously been performed [95–98]. While
the studies had different approaches to the modeling, they were all based on differential equations
describing tumor growth. This approach has the advantage of being semi-physiological in nature, as
it attempts to describe the growth cycle of the tumor. However, the drawback of this approach is
that the drug effect is difficult to interpret as it is often described through kill rates. Furthermore,
the models can be parameter intensive, thereby requiring a large amount of data.
The general pharmacodynamic interaction (GPDI) model, is a relatively new model, which is
considered in the third paper of the thesis. The model uses a semi-mechanistic approach and can
be combined with both Response Additivity and Bliss Independence [99]. The GPDI model identifies
victim and perpetrator drugs in one- or two-way interactions in the potency and/or maximal effect
of the compounds. An example of a one-way interaction in the potency where drug A is the victim
and drug B is the perpetrator is shown in EQ6. The parameters in the model are: 𝐸𝑚𝑎𝑥,𝐴 maximal
effect of A, 𝐶𝐴 concentration of A, 𝐸𝐶50,𝐴 half-maximal concentration of A, 𝐼𝑁𝑇𝑚𝑎𝑥,𝐵→𝐴 maximal
interaction of B on A, 𝐼𝑁𝑇50,𝐵→𝐴 half-maximal interaction concentration, 𝐶𝐵 concentration of B.
𝐸𝐴 =
𝐸𝑚𝑎𝑥,𝐴 ∙ 𝐶𝐴
𝐸𝐶50,𝐴 ∙ (1 +𝐼𝑁𝑇𝑚𝑎𝑥,𝐵→𝐴 ∙ 𝐶𝐵
𝐼𝑁𝑇50,𝐵→𝐴 + 𝐶𝐵) + 𝐶𝐴
(EQ6)
The advantage of the GPDI approach is that the estimated interaction parameters provide a way to
quantify the interactions between the compounds. Identifying interaction in either the maximal
effect or potency leads to greater interpretability of the interaction, which gives a semi-mechanistic
understanding of the drug-drug interaction. Furthermore, identification of perpetrator and victim
drugs reveals the direction of the interaction, which enables the characterization of drug-drug
interaction networks for identifying promising drug combinations.
P a g e | 32
3 Summary of results In this section the three papers that were produced as part of this PhD project are summarised and
discussed.
Papers:
1. Nøhr‐Nielsen A, De Bruin ML, Thomsen M, Pipper CB, Lange T, Bjerrum OJ, Lund TM. Body
of evidence and approaches applied in the clinical development program of fixed‐dose
combinations in the European Union from 2010‐2016. Br J Clin Pharmacol. 2019;(July
2018):1–12.
2. Nøhr‐Nielsen A, Lange T, Forman JL, Papathanasiou T, Foster DJR, Upton RN, Bjerrum OJ,
Lund TM. Demonstrating Contribution of Components of Fixed-Dose Drug Combinations
Through Longitudinal Exposure-Response Analysis. AAPS J 2020 222 22:1–14.
3. Nøhr‐Nielsen A, Bagger SO, Brünner N, Stenvang J, Lund TM. Pharmacodynamic modelling
reveals synergistic interaction between docetaxel and SCO-101 in a docetaxel-resistant
triple negative breast cancer cell line. Eur J Pharm Sci 2020 105315.
P a g e | 33
3.1 Paper 1 - Body of evidence and approaches applied in the clinical
development program of fixed‐dose combinations in the European Union
from 2010‐2016
This study was an analysis of the current state of the clinical development programs of fixed-dose
combinations. Throughout this summary, the terminology from paper 1 will be used to ensure
clarity. The study focused on data from the European Union, specifically the fixed-dose
combinations approved through the central procedure at the EMA. The information regarding the
approved fixed-dose combinations was obtained from the EPARs. Following a range of selection
criteria, outlined in Figure 6, the final pool of fixed-dose combinations included in the analysis was
36. These 36 EPARs included a total of 239 clinical trials and 157.514 patients in their clinical
development programs. The purpose of the study was to characterize the approved fixed-dose
combinations, identify the strategies employed to gain approval, and assess the volume of evidence
in the submissions. An overview of the reviewed EPARs is located in Appendix S2 for paper 1 [5].
Figure 6 – Flowchart showing the identification of excluded and included European Public Assessment Reports (EPARs) in paper 1. Distribution of the included EPARs in the anatomical therapeutic chemical (ATC) classification system is shown to the left. The condition for “special exclusion condition” are outlined in Appendix S1 of paper 1 [5]. The figure and caption are from [5].
P a g e | 34
3.1.1 Characterization of fixed-dose combinations
The first metric analyzed in the study is the anatomical therapeutic chemical (ATC) codes. This
information was extracted from the “Product information and Authorized presentations”
documents. Based on the classification, three major groups of fixed-dose combinations were
identified, metabolic diseases, anti-infectives, and respiratory diseases. This finding corresponds
well with a similar result observed for the FDA, where anti-infectives, metabolic diseases, and
cardiovascular diseases were the therapeutic areas of greatest interest [100]. Reprofiling of drugs
was assessed by comparing the ATC codes of components with the ATC code of the parent
compound. Sharing 0 or 1 level of the ATC code was considered as the drug being reprofiled. A small
group (6%) of fixed-dose combinations was composed of reprofiled drugs. Hence, the predominant
strategy within the development was to improve treatment within the existing therapeutic area.
While small, the group of reprofiled drugs represent an important development path, as it has been
shown that the success rate is higher and expenses lower for reprofiled drugs [101].
From the same documents as ATC codes, the number of approved dose levels for each fixed-dose
combination was extracted. The largest group (58%) of fixed-dose combinations had one approved
dose level, while the remaining fixed-dose combinations were distributed on two, three, and four
dose levels. This data was reviewed in the light of the development of personalized medicines, which
aim to provide treatment tailored to the individual disease [102]. The initial result is that the
inherent limitation of drugs with a fixed drug dose ratio makes fixed-dose combinations and
personalized medicines incompatible. However, for the group of fixed-dose combinations with two
or more dose levels, some level of personalization can be achieved. This flexibility makes the group
particularly important, as personalized medicines are hailed as a transformation of drug
development and clinical use [102, 103].
The substance status assessed the approval state of the components in each fixed-dose
combination. Substance status of the fixed-dose combinations was classified as either two-
approved drugs (AD+AD), one approved and one new molecular entity (AD+NME), or two new
molecular entities (NME+NME). The majority (58%) of the fixed-dose combinations were composed
of two approved drugs. Consequently, 42% of the fixed-dose combinations contained a new
molecular entity. Comparing this distribution to the FDA, where only 25.4% contained a new
molecular entity [100], it seems that there is slightly more focus on including new molecular entities
in fixed-dose combinations in the European Union.
Furthermore, substance status was analyzed for its influence on the body of evidence and the
employed strategies, which is summarized in the following sections.
3.1.2 Body of evidence
The body of evidence was assessed with regards to the number of clinical trials, arms, patients, and
dose levels studied included in the submissions. Furthermore, the data were analyzed with a
generalized linear model to elucidate the influence of factors such as substance status on the body
of evidence.
P a g e | 35
The data showed that significantly more clinical trials are performed, and significantly more patients
are included as part of phase 3 trials than phase 2 trials. For substance status, it was evident that,
in general, there was an increased amount of evidence as part of the submission as one or more
new molecular entities were included in the fixed-dose combination. This increase was the case for
clinical trials, arms, and patients, but interestingly, not for the number of dose levels studied.
The use of PK-PD modeling and the choice of trial design was also considered in relation to the body
of evidence. These factors, as employed strategies, will be discussed further in section 3.1.3.
The use of modeling was extracted from the submission and categorized into three categories: No
modeling, PK modeling, or PK-PD modeling. Neither the use of PK modeling or PK-PD modeling was
found to have had a significant effect on the number of clinical trials, arms, nor patients included in
the submission. However, the use of PK modeling did result in a significant reduction in the number
of dose levels studied. Based on the result, the study found that understanding the
pharmacokinetics of potential drug candidates is essential in selecting the right dose. The lack of
influence on the remaining body of evidence is not in accordance with previously published studies
[79, 104–106]. This deviation from the literature is thought to arise from the lack of details in the
use of modeling in the EPARs. Essentially the modeling was analyzed in a yes/no manner that does
not capture the purposes behind the use of modeling, which in turn might not correspond to the
purposes investigated in this study.
The choice of clinical trial design in the dose findings or main pivotal trial was categorized as either:
Factorial design: Two or more combinations (different ratio)
Ray design: Two or more combinations (same ratio)
Single combination: One combination tested
The influence of clinical trial choice on the body of evidence was significant for the number of
patients and arms. Here, there were significantly more patients and arms when using a factorial
design compared to ray design. Naturally, for the number of doses tested, there was significantly
fewer for single combination compared to factorial design.
3.1.3 Employed strategies in development
Whether dose-finding studies were performed during fixed-dose combination development was
found to be affected by the substance status of the fixed-dose combination. For 57% of the fixed-
dose combinations consisting of two-approved drugs, no dose-finding study was performed. This
approach is described in the relevant guideline for the analyzed period (2009 guidelines) from the
EMA [107], and the present study supports that this approach is possible in practice. Additionally,
the current 2017 guidelines from the EMA includes a section describing the evidence base,
supporting the same approach [51]. A review of the FDA found a similar result, where approval was
granted for fixed-dose combinations without completing the full phases of clinical trials [100].
Utilizing the existing dose levels does, however, have a drawback, as the exploration of the drug-
drug interaction space could have resulted in more optimal dose levels or ratios.
P a g e | 36
Clinical trial design was evaluated to assess the strategy for fulfilling the requirement of
demonstrating the contribution of components to the overall effect. Approximately half (47%) of
the approved fixed-dose combinations had only one dose studied. Furthermore, considering that
Single combination can be considered a subset of Factorial design, the vast majority made use of a
factorial design (47+44%). Notably, the 2009 EMA guidelines suggest the use of multilevel factorial
design, thus, distinguishing between the Single combination and Factorial design groups [107].
Finally, few sponsors made use of the ray design (8%).
Achieving a personalized treatment with a fixed-dose combination requires several approved
combination doses to be available for patients. Consequently, obtaining approval for these
combination doses leads to very large factorial design studies. Thus, the extensive use of the
factorial design within fixed-dose combination development presents a hurdle for personalization
of treatment with fixed-dose combinations. Alternative approaches to the factorial design study
may be found in exposure-response modeling or model-based adaptive optimal design, which could
reduce the need for patients [86, 104]. However, exposure-response analysis of fixed-dose
combinations has been shown to have an inflated false-positive rate, which makes the approach
unfeasible [86]. Conversely, longitudinal exposure-response analysis has been shown to provide a
modeling approach that can reduce the need for patients while producing reliable results [105].
In the study, it was shown that 42% of the clinical development programs included no modeling,
36% performed PK modeling, and 22% performed PK-PD modeling. Employing PK-PD modeling can
utilize the information from early clinical trials to assist in dose selection and provide insights for
expected effect sizes [108]. Furthermore, characterization of the PK profile with covariates, such as
weight and age, is essential, as it can enable extrapolation to special populations [109]. Therefore,
it was surprising that only about half (58%) of the development programs utilized either PK or PK-
PD modeling as a development strategy.
3.1.4 Conclusions
In conclusion, the study emphasizes that interesting approaches are being employed in fixed-dose
combination development, utilizing prior knowledge, not performing dose-finding trials, and
reprofiling of drugs. Personalization of treatment with fixed-dose combinations could be a
promising approach to ensure continued increase in the development of fixed-dose combinations,
however, the extensive use of the factorial design study presents a hurdle for this approach. Lastly,
given the advantages of performing PK-PD modeling and the low use of modeling in the
development of fixed-dose combinations shown in this study, developers of fixed-dose
combinations should to a greater extent consider incorporating the use of modeling in the
development process.
P a g e | 37
3.2 Paper 2 - Demonstrating Contribution of Components of Fixed-Dose Drug
Combinations Through Longitudinal Exposure-Response Analysis In this study, the focus was on exposure-response modeling of fixed-dose combinations. In a
previously published study, it was found that analyzing fixed-dose combinations using exposure-
response modeling caused inflated false positive rates, corresponding to finding an effect of a fixed-
dose combination component when the component had no true effect [86]. This inflation of false
positive rates resulted from confounding factors due to the high correlation between the
concentrations of two components administered in fixed ratios across multiple dose levels [86].
3.2.1 Analysis strategy
This study considered an alternative approach in the form of longitudinal exposure-response
modeling, which utilizes the entire timecourse of the exposure-response relationship. The
hypothesis was that since fixed-dose combinations often have endpoints that have a delay in the
concentration-effect relationship, the advantages of longitudinal exposure-response modeling
could alleviate the issue with inflation of false positives.
The entire study was based on the simulation of clinical trials in which the exposure-response and
longitudinal exposure-response modeling analysis were compared with respect to false positive and
false negative rates. The simulated PK was nominally based on models for empagliflozin (compound
A) and linagliptin (compound B), while the simulated PD was based on their combination Glyxambi®,
which is used for the treatment of T2DM [110–112]. In addition, the influence of the clinical trial
parameters on the false positive and false negative rates was investigated by including 432 scenarios
that varied across the following: Drug activity, number of patients, duration, sampling frequency,
dose distribution, and clinical trial design, Figure 7. Importantly, to evaluate the false positive and
false negative rates, the drug activity of one compound in the fixed-dose combination was fixed to
0. The study was performed in R 3.5.1 using the mrgsolve and lme4 packages for simulation and
analysis, respectively [113–115].
Figure 7 – Flow-chart illustrating the process of the study. Exposure as trough concentrations and response as HbA1c levels are sampled pairwise from the simulated data of each scenario. The samples are then analyzed using exposure-response and longitudinal exposure-response models, and the false positive and false negative rates are computed. Lastly, clinical trial parameters across all scenarios are analyzed. The figure and caption are from paper 2.
3.2.2 Exposure-response analysis
A two-compartment model with first-order absorption and elimination was used to simulate the
exposure, sampled as trough concentrations, while an Imax model with an inhibitory effect on the
formation constant kin constituted the pharmacodynamic model [116].
P a g e | 38
P a g e | 39
Figure 8 – Pharmacodynamic response (HbA1c [%]) vs. compound A (left panel) and B through concentration (middle panel) on a linear and logarithmic scale (embedded graph) as well as PD response vs. time (right panel) across four scenarios. Each scenario includes; 60 subjects, 8 samples, and a trial duration of 100 days (A+B), 30 subjects, 4 samples, and a trial duration of 20 days (C+D). All scenarios use a multiple dosing scheme and a dose distribution on either side of IC50. For scenario A+C, compound A is active, while for B+D, compound B is active. Each data pair is represented by dots colored by dose level. The dashed line represents the loess line, and the solid line reflects the underlying relationship between PD response and compound concentration. The figure and caption are from paper 2.
The exposure-response of four simulated scenarios was investigated in detail and is illustrated in
Figure 8. These scenarios represented what was considered high information and low information
scenarios in the spectrum of scenarios investigated in the study.
Panels A+B are high information scenarios with compound A or B active, respectively. Panels C+D
are low information scenarios with compound A or B active, respectively. From these results, it was
evident that for the active component in the high information scenarios, the loess fit, based on visual
inspection, closely matched that of the true exposure-response relationship from the simulation
model. This was not the case to the same degree for the low information scenarios. However, the
more interesting part was that while the relationships matched for the active component, the
inactive components were more erroneously described for the high information scenario. In
essence, this demonstrated that as the information increased in the clinical trials, the exposure-
response relationship for the active component was better described, but simultaneously, the
inactive component was worse described. This occurred due to confounding, through the high
correlation between concentrations of drug combinations administered in fixed ratios.
A similar result was derived from the exposure-response and longitudinal exposure-response
analysis of false positive and false negative rates. Here the high and low information scenarios were
again considered. For exposure-response analysis, it was clear that the high information clinical
trials lead to an increase in the observed false positive rates compared to the low information clinical
trials. Ultimately, based on these results, performing exposure-response trials for fixed ratio drug
combinations will cause a false claim of contribution of the components to the overall efficacy.
For longitudinal exposure-response, the false positive rates were generally well-controlled across
all scenarios, which essentially showed that longitudinal exposure-response could be used for the
analysis of fixed-dose combinations, without risking false claim of efficacy. Except for the high
information scenario, the false negative rates were generally not well-controlled, especially when
the intercept was fixed to 0 in the analysis model. This is a common approach when it is known that
the data must pass through the origin. However, this approach caused high false negative inflation
for longitudinal exposure-response analysis, which was argued to be caused by an over-simplified
regression model. Another misspecification in the analysis was for the non-linear models with
regards to time, as these models showed an increase in false positive rates in the high information
scenario. In the study, exponential time decay was used for the non-linear description of time [84].
The true description of time is more complex, described here [117], than exponential time decay,
and this simplification may lead to the inflated false positive rates. Another approach that was
considered was a pharmacometric approach, where more mechanistic models could be employed
[118]. However, as the simulation study is based on one of these mechanistic models, performing
P a g e | 40
the analysis with the same model would constitute an artificial best-case scenario. A comparison
between the statistical and pharmacometric approaches could be of interest for future research.
3.2.3 Clinical trial parameters
The clinical trial parameters were explored to provide information on how they affect the false
positive and false negative rates. For exposure-response, the same pattern was found as previously,
showing that more information, through e.g. more patients, led to higher false positive rates. False
positive rates for longitudinal exposure-response was not affected by any parameters,
demonstrating that regardless of scenario, false positive rates are well-controlled. From the clinical
trial parameters, the most important finding was that the inflated false negative rate for longitudinal
exposure-response was caused by the scenarios generally providing too little information. Across all
parameters, false negative rates decreased as more information was included in the trial. Especially
the sequentially administered doses were essential in controlling the false negative rate. In
summation, longitudinal exposure-response analysis showed well-controlled false positive rates
and, given sufficient information corresponding to the high information scenario, also showed well-
controlled false negative rates.
3.2.4 Discussion
The hypothesis in this study was that longitudinal exposure-response analysis could be applied to
fixed-dose combinations without obtaining results with inflated false positive rates. Based on the
results presented in the study, longitudinal exposure-response analysis can be confidently used as
the issues with inflated false positive rates observed in the previous study by Zhu and Wang are
practically eliminated [86]. However, the analysis does have drawbacks. In particular, the sequential
administration of treatment and adequate coverage of the response-time curve were important in
identifying the true exposure-response relationship. In combination, these factors will lead to long
clinical trials, which have higher costs [119]. Conversely, the gain from this analysis is that the
monotherapy arms used in the conventional factorial clinical trial, which is recommended by EMA
[51], becomes redundant. By removing these arms, the clinical trial size can either be reduced, thus
having lower costs, or more patients can be allocated to the combinations, thereby providing more
information on the fixed-dose combination. A general consideration from the study is that the
results apply to any fixed ratio drug combination. However, drug combinations given in a defined
ratio are a common scenario for fixed-dose combinations [120], and the discussion is therefore
focused on fixed-dose combinations.
3.2.5 Conclusion
The conclusion from the study highlights that the previously demonstrated inflated false positive
rate in the exposure-response analysis was not present when the longitudinal exposure-response
analysis was performed. Thus, longitudinal exposure-response analysis can be recommended for
the analysis of fixed-dose combinations or drug combinations in a fixed ratio. It was considered
whether the results from this analysis would carry regulatory weight, concluding, that at least it
would provide supportive information regarding the claim of contribution of each component to
the overall effect. Lastly, carrying out this analysis outside in silico studies would help support the
validity of longitudinal exposure-response analysis of fixed-dose combinations.
P a g e | 41
3.3 Paper 3 - Pharmacodynamic modelling reveals synergistic interaction
between docetaxel and SCO-101 in a docetaxel-resistant triple negative
breast cancer cell line
In the third paper, a drug combination undergoing investigation for its potential use in breast cancer
is analyzed using the GPDI model introduced in section 2.3.4. The combination consists of a novel
compound under investigation, SCO-101, which is hypothesized to enable the treatment of several
drug-resistant cancers, and docetaxel, which is part of the neoadjuvant treatment for TNBC [121].
This paper mainly addresses the modeling and analysis of the data while the cellular, mechanistic,
and assay aspects are planned for a separate publication by the lab conducting the experiments.
SCO-101 is hypothesized to enable treatment in drug-resistant cancers through inhibition of the
ATP-binding cassette ABCG2, which is a human multidrug transporter [122, 123]. The protein is
responsible for the efflux of commonly used pharmaceuticals and thereby contributes to drug
resistance in cancers [123]. Furthermore, a phase II clinical study of SCO-101 in combination with
FOLFIRI is planned in patients with colorectal cancer using the same rationale [124]. FOLFIRI is an
established regiment for the treatment of colorectal cancer [124].
The development of drug resistance is one of the primary barriers in treating patients suffering from
cancer [14, 125]. In theory, targeting drug efflux pumps enables several potential substrates. TNBC
is outlined in section 2.2.2 and represents a cancer subtype with a very poor prognosis and a high
degree of drug resistance. This makes TNBC a good candidate for investigating potential treatment
improvement through the addition of SCO-101 to the existing regiment. Therefore, the effects of
SCO-101, docetaxel, and the combination was investigated in a docetaxel resistant TNBC cell line,
MDA-MB-231.
Pharmacodynamic modeling was conducted as an alternative to more conventional approaches,
such as analysis using t-tests or ANOVA. The hypothesis here was that more information could be
extracted from the data by employing the GPDI model, which provides a semi-mechanistic
understanding of the combination, as opposed to considering whether there are significant
differences in cell survival between groups.
3.3.1 Monotherapy models
The initial strategy was to establish the concentration-effect relationship for docetaxel and SCO-101
administered separately. The four relationships, linear, log-linear, Imax and sigmoidal Imax in EQ7 was
considered for both drugs. The parameters in the models are outlined in section 2.3.2. The
relationship for docetaxel was best described by an Imax model, while the best description was
achieved for SCO-101 with a sigmoid Imax model based on the OFV. The model fits plotted on top of
the observed data from the experiments can be seen in Figure 9.
𝐼𝑑𝑟𝑢𝑔 = α ∙ 𝐶𝑑𝑟𝑢𝑔 + β ; 𝐼𝑑𝑟𝑢𝑔 = α ∙ log (𝐶𝑑𝑟𝑢𝑔) + β
𝐼𝑑𝑟𝑢𝑔 =𝐼𝑚𝑎𝑥,𝑑𝑟𝑢𝑔∙𝐶𝑑𝑟𝑢𝑔
𝐼𝐶50,𝑑𝑟𝑢𝑔+𝐶𝑑𝑟𝑢𝑔 ; 𝐼𝑑𝑟𝑢𝑔 =
𝐼𝑚𝑎𝑥,𝑑𝑟𝑢𝑔∙𝐶𝑑𝑟𝑢𝑔
𝐻𝑑𝑟𝑢𝑔
𝐼𝐶50,𝑑𝑟𝑢𝑔
𝐻𝑑𝑟𝑢𝑔+𝐶𝑑𝑟𝑢𝑔
𝐻𝑑𝑟𝑢𝑔
(EQ7)
P a g e | 42
Figure 9 – MDA-MD-231 survival following monotherapy with docetaxel (A+B) or SCO-101 (C+D). The blue dots represent the samples from the experiment, while the black line corresponds to average cell survival. The red dashed line represents the fitted curve for docetaxel (Imax model) and SCO-101 (sigmoid Imax model). B and D contain the same data as A and C, respectively, but on a logarithmic scale. The figure and caption are from paper 3 [126].
For the data presented in Figure 9, the y-axis represents cell survival measured as optical density,
where ~0.3 OD corresponds to ~100% cell survival. The model parameters showed that the two
compounds had similar maximal effects, albeit with less certainty of the estimate for SCO-101. The
maximal effect for docetaxel was around 80%, with a half-maximal inhibitory concentration of 0.413
µM. These parameter values were in agreement with a previous study of docetaxel in docetaxel
resistant MDA-MB-231 cells [127]. The cytotoxicity of SCO-101 alone could be attributed to the
inhibition of SRPK1 kinase, which is involved in tumor growth [124].
3.3.2 Combination model
Following the development of the monotherapy models, several combination models were
evaluated. These included Response Additivity [20], Bliss Independence [92], and finally, a general
pharmacodynamic interaction model [99] of the best fit between the two. Bliss Independence
provided the better model fit and was used as the basis for the GPDI model, which is outlined in
EQ8.
𝐼𝐷𝑜𝑐𝑒 =𝐼𝑚𝑎𝑥,𝐷𝑜𝑐𝑒 ∙ 𝐶𝐷𝑜𝑐𝑒
𝐼𝐶50,𝐷𝑜𝑐𝑒 ∙ (1 +𝐼𝑁𝑇𝑚𝑎𝑥,𝑆𝐶𝑂→𝐷𝑜𝑐𝑒 ∙ 𝐶𝑆𝐶𝑂
𝐼𝑁𝑇50,𝑆𝐶𝑂→𝐷𝑜𝑐𝑒 + 𝐶𝑆𝐶𝑂) + 𝐶𝐷𝑜𝑐𝑒
; 𝐼𝑆𝐶𝑂 =𝐼𝑚𝑎𝑥,𝑆𝐶𝑂 ∙ 𝐶𝑑𝑟𝑢𝑔
𝐻𝑑𝑟𝑢𝑔
𝐼𝐶50,𝑑𝑟𝑢𝑔
𝐻𝑑𝑟𝑢𝑔 + 𝐶𝑑𝑟𝑢𝑔
𝐻𝑑𝑟𝑢𝑔
(EQ8)
P a g e | 43
Other variants of the GPDI model were investigated with a 1-way interaction in the opposite
direction of EQ8 and with a 2-way interaction between the drugs. However, these models
encountered boundary errors and issues with minimization even with several fixed parameters.
Therefore, the model in EQ8 was considered to provide the best fit.
The most interesting parameters extracted from the model were the 𝐼𝑁𝑇𝑚𝑎𝑥,𝑆𝐶𝑂→𝐷𝑜𝑐𝑒 designating
the maximal interaction effect of SCO-101 on the potency of docetaxel, and 𝐼𝑁𝑇50,𝑆𝐶𝑂→𝐷𝑜𝑐𝑒 the
concentration of SCO-101 for half-maximal interaction with docetaxel. The maximal interaction
𝐼𝑁𝑇𝑚𝑎𝑥,𝑆𝐶𝑂→𝐷𝑜𝑐𝑒 was estimated to -0.604, which can be interpreted as an approximately 60%
increase in the potency of docetaxel at maximum interaction effect when SCO-101 is administered
with docetaxel. The half-maximal concentration of the interaction 𝐼𝑁𝑇50,𝑆𝐶𝑂→𝐷𝑜𝑐𝑒 was estimated to
30.9 µM, which is approximately half of the IC50 for the compound itself, which is 59.4 µM.
Figure 10 – Contour plot of the final general pharmacodynamic interaction model. The lines indicate four separate response levels of 0.1 OD, 0.07 OD, 0.05 OD, and 0.03 OD across the dose combination space. 0.05 OD corresponds to the 85% reduction in cell viability target. The four dots highlight dose pairs that result in meeting this target, through lowest total dose combination (red), minimized exposure to both compounds (green), and with a weighted penalty factor of 2 on lowest total dose combination (orange) and minimized exposure (yellow). The figure and caption are from paper 3 [126].
P a g e | 44
The model was used to provide information on dose ratios for further studies. These estimations
were based on locating the lowest total drug combination and the minimized exposure to both
compounds. The former of the two methods was based on the addition of scaled doses, and the
latter was based on Pythagoras’ theorem. Lastly, as docetaxel represented the more toxic of the
two compounds, a penalty was considered. The resulting contour plot from the model and the
optimal dose ratios are illustrated in Figure 10.
The optimal dose ratios were based on the selected effect target of 85% reduction in cell survival
and resulted in dose ratios between 1:40 and 1:64 (docetaxel:SCO-101) being promising for further
analysis.
3.3.3 Discussion
Preclinical modeling of cancer therapeutics has previously been performed as outlined in section
2.3.4. Some of the core issues that arise from these models are the lack of interpretability of the
results and selection of an additivity criterion. The lack of interpretability in these models stems
from describing the drug effect through kill rates. Here the empirical nature of the GPDI model
provides results expressed as maximal effects and half-maximal concentrations, which can be
considered more intuitive. Selecting an additivity criterion causes the classification of a combination
as either synergistic or antagonistic to be dependent on assumptions tied to the specific additivity
criterion. As an example, selecting Bliss Independence includes base assumptions of the underlying
mechanism of the combined drugs, i.e. that they act independently [91, 92]. The GPDI model unifies
the interpretation of these additivity criteria in a model-based framework, which addresses the
issues with selecting an additivity criterion [99].
A simpler approach would be to employ a more conventional analysis such as t-tests or ANOVA.
These methods avoid some of the limitations of the modeling approach, such as selecting an
additivity criterion, establishing a model, and the interpretability of the results. However, as
highlighted in the following paragraphs, using models such as the GPDI model provides significant
insights into drug-drug combinations that go far beyond a comparison of groups.
The potency of the interaction 𝐼𝑁𝑇50,𝑆𝐶𝑂→𝐷𝑜𝑐𝑒 was estimated to 30.9 µM. This corresponds to half
the concentration needed to reach the half-maximal effect on cell survival of SCO-101, which has
an IC50 of 59.4 µM. The essence of this discovery was that at concentrations near the half-maximal
concentration of the interaction, the effect of SCO-101 is primarily mediated through its interaction
with docetaxel. Thus, depending on tolerability for SCO-101 the primary treatment capacity of SCO-
101 will be its interaction with docetaxel. Usually, the underlying implicit assumption is that the
drug effect and interaction share the same potency [127]. Estimating the potency of the interaction
is a key trait of the GPDI model, which provided important information on the difference between
the potency of SCO-101 and the potency of its interaction with docetaxel.
The maximal interaction effect 𝐼𝑁𝑇𝑚𝑎𝑥,𝑆𝐶𝑂→𝐷𝑜𝑐𝑒 was estimated to -0.604. This corresponds to a 60%
reduction in the half-maximal concentration of docetaxel when administered in combination with
SCO-101 compared to when docetaxel is administered alone. By estimating the maximal interaction,
the GPDI model provides a semi-mechanistic understanding of the interaction. Two clinical
P a g e | 45
implications arise from this parameter. Either the dose of docetaxel can be reduced when given in
combination with SCO-101, reducing toxicity but maintaining the same effect, or the dose of
docetaxel can be maintained, thus attaining a higher effect with no additional side effects from
docetaxel. Identifying victim and perpetrator drugs in the combination is an implicit part of the GPDI
model, which enables the identification of interaction networks for large combination analyses [99].
Overall, analysis of the combination data using modeling provided significant insights over
conventional approaches such as t-tests or ANOVA.
The clinical relevance of the doses for both SCO-101 and docetaxel should also be considered. SCO-
101 was found to have limited toxicity when administered alone, indicating that there will be few
potential issues with the doses [122, 128]. However, to the knowledge of the authors, no concrete
information on the plasma concentration of SCO-101 is available, thus, evaluation of the clinical
relevance of the SCO-101 concentration can be considered speculation. Conversely, the
pharmacokinetics of docetaxel are well documented. A pharmacokinetic analysis of docetaxel found
the median Cmax to be 3.7 µM and the range to span 2.6-6.9 µM in patients receiving 100 mg/m2
docetaxel [129]. These values correspond well with the investigated range of docetaxel
concentrations in the present study making them relevant from a clinical perspective. Furthermore,
the identified optimal dose pairs in the study all recommend doses of docetaxel below the median
Cmax of 3.7 µM. Thus, the recommendations for the drug-drug ratio of SCO-101 and docetaxel seem
sensible.
Assessing the potential of a drug combination in humans, based on in vitro experiments, is
challenging. The increase in complexity from in vitro to in vivo is immense. The effects of aspects
such as ADME and cellular aspects such as the immune system make extrapolation unreliable. One
approach that can assist in utilizing in vitro data for extrapolation to in vivo is in vitro in vivo
extrapolation (IVIVE). IVIVE makes use of modeling and simulation to make quantitative
extrapolation of drug exposures [130, 131]. The major determinants for this method is establishing
a physiologically based model and carrying out in vitro studies of solubility in GI fluid and
permeability [130]. Overall, this presents a fairly complex process but could provide a method for
extrapolating the combination data. However, realistically, to make confident conclusions about the
combination in humans, studies in vivo and in humans will be required.
3.3.4 Conclusion
In conclusion, a model describing the pharmacodynamics of the combination of SCO-101 and
docetaxel was established. Based on the information from this model, it can be concluded that the
novel compound SCO-101 has the potential to provide a significant improvement in the treatment
of docetaxel resistant TNBC when administered in combination with docetaxel. Modeling the in vitro
data provided significant insights into the combination compared to conventional approaches
through estimation of the maximal interaction and the potency of the interaction. Furthermore, the
information from the model enabled recommendations on the optimal drug-drug ratio. Lastly, the
study serves as a case study of the GPDI model, which was shown to offer significant advantages
over conventional methods.
P a g e | 46
4 Conclusions and Perspectives for further research Based on the research conducted in this PhD thesis, three papers were published:
1. Nøhr‐Nielsen A, De Bruin ML, Thomsen M, Pipper CB, Lange T, Bjerrum OJ, Lund TM. Body
of evidence and approaches applied in the clinical development program of fixed‐dose
combinations in the European Union from 2010‐2016. Br J Clin Pharmacol. 2019;(July
2018):1–12.
2. Nøhr‐Nielsen A, Lange T, Forman JL, Papathanasiou T, Foster DJR, Upton RN, Bjerrum OJ,
Lund TM. Demonstrating Contribution of Components of Fixed-Dose Drug Combinations
Through Longitudinal Exposure-Response Analysis. AAPS J 2020 222 22:1–14.
3. Nøhr‐Nielsen A, Bagger SO, Brünner N, Stenvang J, Lund TM. Pharmacodynamic modelling
reveals synergistic interaction between docetaxel and SCO-101 in a docetaxel-resistant triple
negative breast cancer cell line. Eur J Pharm Sci 2020 105315.
The studies covered in each of the three papers addressed the overall aim and the three objectives
outlined in section 1.
The first objective was to assess the current practice of the use of modeling as a drug development
tool for the development of fixed-dose combinations in the European Union. This was achieved in
paper 1: “Body of evidence and approaches applied in the clinical development program of fixed‐
dose combinations in the European Union from 2010‐2016”. In the study, it was shown that only
58% of approved fixed-dose combinations made use of PK or PK-PD modeling during development.
Given the advantages outlined in the discussion of paper 1 (section 3.1.3), this was surprising. The
main application of this result is that developers of fixed-dose combinations should to a greater
extent consider incorporating the use of modeling in the development process. Other novel results
from the study include that:
Performing PK modeling lead to significantly fewer doses investigated
Components of the fixed-dose combinations are to a small degree reprofiled (6%)
Fixed-dose combinations approved by EMA primarily consist of two previously approved
drugs (71%) and have a single approved combination dose (71%)
No dose-finding trial was conducted for more than half of fixed-dose combinations
composed of two previously approved drugs (57%)
The main limitation of the study, for achieving the first objective in the thesis, was that the EPARs
contain no information on the actual models or the purpose for employing them. Therefore, the
study assessed the use of modeling in a yes/no manner. This limitation was highlighted by the fact
that PK-PD modeling was shown to have no impact on the number of patients, arms, clinical trials
or doses tested, which is not in agreement with previously published studies [79, 104–106].
The continuance of this research would need to address this limitation by analyzing more detailed
information regarding the modeling conducted for the development of fixed-dose combinations.
Particularly with a focus on the type of model and the purpose for employing it.
P a g e | 47
In summation, the first objective of the thesis was fulfilled, however, more in-depth knowledge on
the use of modeling as a drug development tool for the development of fixed-dose combinations in
the European Union can be gained by performing the suggested studies.
The second objective was to develop new methods that may assist in the drug development process
of fixed-dose combinations. This was achieved in paper 2: “Demonstrating Contribution of
Components of Fixed-Dose Drug Combinations Through Longitudinal Exposure-Response Analysis”.
In this in silico study, the applicability of performing longitudinal exposure-response modeling in the
development of fixed-dose combinations was shown. Utilizing longitudinal exposure-response in
this setting constituted a new method for demonstrating the contribution of the components to the
overall effect without an inflated false positive rate, which had been present for exposure-response
analysis in a previous study [86].
The application of this research is that the monotherapy arms used in the conventional factorial
clinical trial can be eliminated as they are not required for this type of analysis. The gains from
removing these arms are either allocation of more patients to the combination arms or reduction
in clinical trial size. Thus, there is either an increase in information on the combination or a reduction
in cost and the number of patients needed to treat. Additional novel results included the influence
of several parameters on the false positive and false negative rate in both exposure-response and
longitudinal exposure-response analysis. In particular, the sequential administration of the doses in
the clinical trial and adequate coverage of the response-time curve was essential in identifying the
true exposure-response relationship.
The primary limitation of the findings in this study originate from the in silico setting. This setting
has overarching assumptions about the parameter choices for both the simulated compounds and
clinical trials. The most important assumption was that the chosen parameters reflect a realistic
setting. To expand upon this research, it will be important to apply longitudinal exposure-response
modeling to data from previously conducted clinical trials and assess if the results match those from
the original methods. Additionally, defining the longitudinal exposure-response model a priori and
analyzing clinical trial data will verify the applicability of the method.
Based on the research in paper 2 the second objective has been achieved. While longitudinal
exposure-response modeling is not in itself a new method, the application of the method for fixed-
dose combinations is novel and will aid in the development of fixed-dose combinations.
The third objective was to validate combination models on preclinical data of a combination under
investigation and assessing the value of applying these models in guiding further investigations of
the combination. This was achieved in paper 3: “Pharmacodynamic modelling reveals synergistic
interaction between docetaxel and SCO-101 in a docetaxel-resistant triple negative breast cancer
cell line”.
In this in vitro study, a combination of the novel compound SCO-101 and docetaxel was used for the
treatment of docetaxel resistant TNBC and analyzed using PD modeling. Several models were
evaluated and the best model fit was found to be the GPDI model. Based on this model, it was found
that SCO-101 interacted with docetaxel, leading to a maximal reduction in the IC50 value of docetaxel
by 60%. Therefore, the combination was characterized as synergistic. Analysis of the model
P a g e | 48
identified optimal drug-drug ratios between 1:40 and 1:64 (docetaxel:SCO-101). Furthermore, the
study also presents a case study of the GPDI model, which is a fairly recently developed model [99].
The application of these results is that the combination of SCO-101 and docetaxel has the potential
to provide a significant improvement in the treatment of docetaxel resistant TNBC and is therefore
worth further investigation. Furthermore, the research highlights that the parameters from the
GPDI model led to a greater understanding of the drug-drug interaction than conventional
approaches and provided valuable information to guide further investigations through dose ratio
selection.
A key limitation of modeling in vitro studies is the very controlled setting that it represents. The
established model showed very low uncertainty in the parameters, which is unfeasible in more
variable settings such as in vivo. Furthermore, cellular aspects such as the presence of the immune
system could potentially influence the efficacy of the combination. Thus, to provide confident
conclusions about the application of the combination in humans, more studies of the combination
in vivo and in clinical trials will be necessary.
Combination models were validated on preclinical data of a combination under investigation and
the value of the model in guiding further investigations was assessed in paper 3. Based on this, the
third objective of this thesis was achieved.
The work of this PhD thesis and previous works highlights both the pitfalls and potential advantages
of utilizing model-based approaches for the development of fixed-dose combinations. The overall
aim of this thesis was accomplished as the use of several model-based approaches has been
explored, developed, and demonstrated within the field of fixed-dose combinations and the
development of these products. Based on the research conducted in this thesis, the overall
recommendation is that modeling tools should to a greater degree be incorporated in the
development of fixed-dose combinations as they assist in determining efficacy and provide valuable
information to guide drug development.
P a g e | 49
5 References 1. Ascierto PA, Marincola FM (2011) Combination therapy: The next opportunity and challenge
of medicine. J Transl Med 9:115. https://doi.org/10.1186/1479-5876-9-115
2. Woodcock J, Griffin JP, Behrman RE (2011) Development of Novel Combination Therapies. N Engl J Med 364:985–987. https://doi.org/10.1056/NEJMp1101548
3. Editorial (2017) Rationalizing combination therapies. Nat Med 23:1113–1113. https://doi.org/10.1038/nm.4426
4. FDA/CDER Hypertension: Developing Fixed-Combination Drug Products for Treatment Guidance for Industry. https://www.fdanews.com/ext/resources/files/2018/2/11-06-18-Hypertension.pdf?1541542345. Accessed 1 Dec 2019
5. Nøhr-Nielsen A, De Bruin ML, Thomsen M, et al (2019) Body of evidence and approaches applied in the clinical development programme of fixed-dose combinations in the European Union from 2010 to 2016. Br J Clin Pharmacol 85:1829–1840. https://doi.org/10.1111/bcp.13986
6. EMA Modelling and Simulation Working Party | European Medicines Agency. https://www.ema.europa.eu/en/committees/working-parties-other-groups/chmp/modelling-simulation-working-party. Accessed 1 Dec 2019
7. EMA/CHMP (2018) Guideline on the reporting of physiologically based pharmacokinetic (PBPK) modelling and simulation. www.ema.europa.eu/contact. Accessed 28 Nov 2019
8. FDA (2019) Population Pharmacokinetics Guidance for Industry (Draft Guidance). https://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.htm. Accessed 9 Dec 2019
9. Morrison T, FDA (2018) How Simulation Can Transform Regulatory Pathways. https://www.fda.gov/ScienceResearch/AboutScienceResearchatFDA/ucm616822.htm
10. Bulusu KC, Guha R, Mason DJ, et al (2016) Modelling of compound combination effects and applications to efficacy and toxicity: State-of-the-art, challenges and perspectives. Drug Discov. Today 21:225–238
11. Hao J, Rodriguez-Monguio R, Seoane-Vazquez E (2015) Fixed-dose combination drug approvals, patents and market exclusivities compared to single active ingredient pharmaceuticals. PLoS One 10:1–13. https://doi.org/10.1371/journal.pone.0140708
12. Hao J, Rodriguez-Monguio R, Seoane-Vazquez E (2016) Fixed-dose combination and single active ingredient drugs: A comparative cost analysis. Expert Rev Pharmacoeconomics Outcomes Res 16:127–134. https://doi.org/10.1586/14737167.2015.1068690
13. Kwon KC, Lee C (2017) Analysis of Fixed-Dose Combination Products Approved by the US Food and Drug Administration, 2010-2015. Ther Innov Regul Sci 51:111–117. https://doi.org/10.1177/2168479016663263
14. Housman G, Snyder N, Byler S, et al (2014) Drug Resistance in Cancer: An Overview. Cancers (Basel) 6:1769–1792. https://doi.org/10.3390/cancers6031769
15. Sawicki-Wrzask D, Thomsen M, Bjerrum OJ (2015) An Analysis of the Fixed-Dose
P a g e | 50
Combinations Authorized by the European Union, 2009-2014. Ther Innov Regul Sci 49:553–559. https://doi.org/10.1177/2168479014567322
16. Kikuchi C, Ohno M, Izumo T, et al (2019) Investigation of Approval Trends and Benefits of New Fixed-Dose Combination Drugs in Japan. Ther Innov Regul Sci 216847901882191. https://doi.org/10.1177/2168479018821919
17. Kitano H (2007) A robustness-based approach to systems-oriented drug design. Nat Rev Drug Discov 6:202–210. https://doi.org/10.1038/nrd2195
18. Yap TA, Omlin A, de Bono JS (2013) Development of Therapeutic Combinations Targeting Major Cancer Signaling Pathways. J Clin Oncol 31:1592–1605. https://doi.org/10.1200/JCO.2011.37.6418
19. Keith CT, Borisy AA, Stockwell BR (2005) Multicomponent therapeutics for networked systems. Nat Rev Drug Discov 4:71–78. https://doi.org/10.1038/nrd1609
20. Foucquier J, Guedj M (2015) Analysis of drug combinations: current methodological landscape. Pharmacol Res Perspect 3:e00149. https://doi.org/10.1002/prp2.149
21. Rekić D, Reynolds KS, Zhao P, et al (2017) Clinical Drug–Drug Interaction Evaluations to Inform Drug Use and Enable Drug Access. J Pharm Sci 106:2214–2218. https://doi.org/10.1016/j.xphs.2017.04.016
22. Navarro J, Curran A (2016) Profile of once-daily darunavir/cobicistat fixeddose combination for the treatment of HIV/AIDS. HIV/AIDS - Res Palliat Care 8:175–182. https://doi.org/10.2147/HIV.S56158
23. Coiffier B, Sarkozy C (2016) Diffuse large B-cell lymphoma: R-CHOP failure—what to do? Hematology 2016:366–378. https://doi.org/10.1182/asheducation-2016.1.366
24. Palmer AC, Chidley C, Sorger PK (2019) A curative combination cancer therapy achieves high fractional cell killing through low cross-resistance and drug additivity. Elife 8:e50036. https://doi.org/10.7554/eLife.50036
25. Woo V (2015) Empagliflozin/linagliptin single-tablet combination: First-in-class treatment option. Int J Clin Pract 69:1427–1437. https://doi.org/10.1111/ijcp.12720
26. Daniel TM (2006) The history of tuberculosis. Respir Med. https://doi.org/10.1016/j.rmed.2006.08.006
27. Liu X, Ma Q, Zhang F (2010) Therapeutic drug monitoring in highly active antiretroviral therapy. Expert Opin Drug Saf 9:743–758. https://doi.org/10.1517/14740331003767395
28. Friedman HS, Rajagopalan S, Barnes JP, Roseman H (2011) Combination Therapy With Ezetimibe/Simvastatin Versus Statin Monotherapy for Low-Density Lipoprotein Cholesterol Reduction and Goal Attainment in a Real-World Clinical Setting. Clin Ther 33:212–224. https://doi.org/10.1016/j.clinthera.2011.02.011
29. WHO (2019) World Health Organization Model List of Essential Medicines. https://apps.who.int/iris/bitstream/handle/10665/325771/WHO-MVP-EMP-IAU-2019.06-eng.pdf?ua=1. Accessed 23 Apr 2020
30. Wertheimer a I (2013) The economics of polypharmacology: fixed dose combinations and
P a g e | 51
drug cocktails. Curr Med Chem 20:1635–8. https://doi.org/10.2174/0929867311320130003
31. Miranda MRH, Dubey A, G.S R, Charyulu RN (2019) Fixed-dose combinations banned in India: is it the right decision? An eye-opening review. Expert Opin Drug Saf 18:977–985. https://doi.org/10.1080/14740338.2019.1651292
32. Gupta Y, Ramachandran S (2016) Fixed dose drug combinations: Issues and challenges in India. Indian J Pharmacol 48:347. https://doi.org/10.4103/0253-7613.186200
33. Bell DSH (2013) Combine and conquer: advantages and disadvantages of fixed-dose combination therapy. Diabetes, Obes Metab 15:291–300. https://doi.org/10.1111/dom.12015
34. Du L-P, Cheng Z-W, Zhang Y-X, et al (2018) The impact of fixed-dose combination versus free-equivalent combination therapies on adherence for hypertension: a meta-analysis. J Clin Hypertens 20:902–907. https://doi.org/10.1111/jch.13272
35. Hutchins V, Zhang B, Fleurence RL, et al (2011) A systematic review of adherence, treatment satisfaction and costs, in fixed-dose combination regimens in type 2 diabetes. Curr Med Res Opin 27:1157–1168. https://doi.org/10.1185/03007995.2011.570745
36. Benford M, Milligan G, Pike J, et al (2012) Fixed-dose combination antidiabetic therapy: Real-world factors associated with prescribing choices and relationship with patient satisfaction and compliance. Adv Ther 29:26–40. https://doi.org/10.1007/s12325-011-0096-z
37. Kawalec P, Holko P, Gawin M, Pilc A (2018) Systematic review/Meta-analysis Effectiveness of fixed-dose combination therapy in hypertension: systematic review and meta-analysis. Arch Med Sci 14:1125–1136. https://doi.org/10.5114/aoms.2018.77561
38. Bahiru E, de Cates AN, Farr MRB, et al (2017) Fixed-dose combination therapy for the prevention of atherosclerotic cardiovascular diseases. Cochrane Database Syst Rev 2017:CD009868. https://doi.org/10.1002/14651858.CD009868.pub3
39. Upcounsel. How Long Does a Drug Patent Last? https://www.upcounsel.com/how-long-does-a-drug-patent-last. Accessed 5 Apr 2020
40. Wertheimer a I (2013) The economics of polypharmacology: fixed dose combinations and drug cocktails. Curr Med Chem 20:1635–8. https://doi.org/10.2174/0929867311320130003
41. Bjerrum OJ, Eichendorff S, Alkis NB (2019) Analysis of the Authorized Target Populations for Fixed Dose Combination Products Between 2000 and 2017 Reveals Discrepancies Between EMA’s and FDA’s Views on Initial Dual-Therapy. Ther Innov Regul Sci 1–6. https://doi.org/10.1177/2168479019836976
42. Herrick TM, Million RP (2007) Tapping the potential of fixed-dose combinations. Nat Rev Drug Discov 6:513–514. https://doi.org/10.1038/nrd2334
43. Times TNY (2006) Heart Pill to Be Sold by Itself. https://www.nytimes.com/2006/07/26/business/26drug.html. Accessed 11 Dec 2019
44. EMA Pricing and reimbursement of medicinal products | Internal Market, Industry, Entrepreneurship and SMEs. https://ec.europa.eu/growth/sectors/healthcare/competitiveness/products-pricing-
P a g e | 52
reimbursement_en. Accessed 1 Apr 2020
45. Melikian C, White TJ, Vanderplas A, et al (2002) Adherence to oral antidiabetic therapy in a managed care organization: A comparison of monotherapy, combination therapy, and fixed-dose combination therapy. Clin Ther 24:460–467. https://doi.org/10.1016/S0149-2918(02)85047-0
46. Rabbani A, Alexander GC (2008) Out-of-pocket and total costs of fixed-dose combination antihypertensives and their components. Am J Hypertens 21:509–513. https://doi.org/10.1038/ajh.2008.31
47. Malesker MA, Hilleman DE (2010) Comparison of amlodipine/valsartan fixed-dose combination therapy and conventional therapy. Manag Care 19:36–42
48. Hong SH, Wang J, Tang J (2013) Dynamic view on affordability of fixed-dose combination antihypertensive drug therapy. Am J Hypertens 26:879–887. https://doi.org/10.1093/ajh/hpt035
49. Gibson TB, Song X, Alemayehu B, et al (2010) Cost sharing, adherence, and health outcomes in patients with diabetes. Am J Manag Care 16:589–600
50. ICH. ICH Mission: Harmonisation for Better Health. https://www.ich.org/page/mission. Accessed 1 Apr 2020
51. EMA/CHMP (2017) Guideline on Clinical Development of Fixed Combination Medicinal Products. In: Eur. Med. Agency EMA/CHMP/158268/2017. https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-clinical-development-fixed-combination-medicinal-products-revision-2_en.pdf. Accessed 28 Nov 2019
52. FDA/CDER (2006) Fixed Dose Combinations, Co-Packaged Drug Products, and Single-Entity Versions of Previously Approved Antiretrovirals for the Treatment of HIV. In: Food Drug Adm. Rockville, MD. https://www.fda.gov/media/72248/download. Accessed 28 Nov 2019
53. Lodola A (2017) Nonclinical Development of Combination Drugs. In: Methods in Molecular Biology. Humana Press Inc., pp 3–24
54. Bjerrum OJ, Gautam Y, Bjerrum EJ, et al (2013) Medicines combinations options and regulatory hurdles. Eur J Pharm Sci 49:659–663. https://doi.org/10.1016/j.ejps.2013.04.028
55. FDA/CDER (2013) Guidance for Industry Codevelopment of Two or More New Investigational Drugs for Use in Combination Clinical Medical Guidance for Industry Codevelopment of Two or More New Investigational Drugs for Use in Combination Clinical Medical Contains Nonbinding R. http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.htm. Accessed 1 Dec 2019
56. WHO (2005) Guidelines for registration of fixed-dose combination medicinal products. In: WHO Tech. Rep. Ser. http://apps.who.int/prequal/info_general/documents/TRS929/WHO_TRS_929_annex5FDCs.pdf
57. Kota J, Ayalavajjala PS, Sivasubramanian R (2015) Development of orally administered fixed
P a g e | 53
dose combination (FDC) products: Pharmacokinetic & biopharmaceutical considerations. 6:3161–3173. https://doi.org/10.13040/IJPSR.0975-8232.6(8).3161-73
58. Prueksaritanont T, Chu X, Gibson C, et al (2013) Drug-drug interaction studies: Regulatory guidance and an industry perspective. AAPS J 15:629–645. https://doi.org/10.1208/s12248-013-9470-x
59. FDA (1999) eCFR — Code of Federal Regulations. https://www.ecfr.gov/cgi-bin/text-idx?SID=487f084485a51d667d431233c1c6028b&mc=true&node=se21.5.300_150&rgn=div8. Accessed 3 Dec 2019
60. Mitra A, Wu Y (2012) Challenges and opportunities in achieving bioequivalence for fixed-dose combination products. AAPS J 14:646–655. https://doi.org/10.1208/s12248-012-9378-x
61. EMA/CHMP European public assessment report | European Medicines Agency. https://www.ema.europa.eu/en/glossary/european-public-assessment-report. Accessed 4 Dec 2019
62. DeFronzo RA, Ferrannini E, Groop L, et al (2015) Type 2 diabetes mellitus. Nat Rev Dis Prim 1:1–23. https://doi.org/10.1038/nrdp.2015.19
63. Defronzo RA (2009) From the triumvirate to the ominous octet: A new paradigm for the treatment of type 2 diabetes mellitus. Diabetes 58:773–795. https://doi.org/10.2337/db09-9028
64. Abdul-Ghani MA, Tripathy D, DeFronzo RA (2006) Contributions of β-cell dysfunction and insulin resistance to the pathogenesis of impaired glucose tolerance and impaired fasting glucose. Diabetes Care 29:1130–1139. https://doi.org/10.2337/dc05-2179
65. Stratton IM, Adler AI, Neil HAW, et al (2000) Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): Prospective observational study. Br Med J 321:405–412. https://doi.org/10.1136/bmj.321.7258.405
66. Cahn A, Cefalu WT (2016) Clinical considerations for use of initial combination therapy in type 2 diabetes. Diabetes Care 39:S137–S145. https://doi.org/10.2337/dcS15-3007
67. Casagrande SS, Fradkin JE, Saydah SH, et al (2013) The prevalence of meeting A1C, blood pressure, and LDL goals among people with diabetes, 1988-2010. Diabetes Care 36:2271–2279. https://doi.org/10.2337/dc12-2258
68. Bouchoucha M, Uzzan B, Cohen R (2011) Metformin and digestive disorders. Diabetes Metab 37:90–96. https://doi.org/10.1016/j.diabet.2010.11.002
69. Marra A, Viale G, Curigliano G (2019) Recent advances in triple negative breast cancer: The immunotherapy era. BMC Med 17:1–9. https://doi.org/10.1186/s12916-019-1326-5
70. Lin NU, Vanderplas A, Hughes ME, et al (2012) Clinicopathologic features, patterns of recurrence, and survival among women with triple-negative breast cancer in the National Comprehensive Cancer Network. Cancer 118:5463–5472. https://doi.org/10.1002/cncr.27581
71. Mehanna J, Haddad FGH, Eid R, et al (2019) Triple-negative breast cancer: Current perspective on the evolving therapeutic landscape. Int J Womens Health 11:431–437.
P a g e | 54
https://doi.org/10.2147/IJWH.S178349
72. Bergin ART, Loi S (2019) Triple-negative breast cancer: Recent treatment advances [version 1; peer review: 2 approved]. F1000Research 8:1–11. https://doi.org/10.12688/f1000research.18888.1
73. Omarini C, Guaitoli G, Pipitone S, et al (2018) Neoadjuvant treatments in triple-negative breast cancer patients: Where we are now and where we are going. Cancer Manag Res 10:91–103. https://doi.org/10.2147/CMAR.S146658
74. Mokhtari RB, Homayouni TS, Baluch N, et al (2017) Combination therapy in combating cancer. Oncotarget 8:38022–38043. https://doi.org/10.18632/oncotarget.16723
75. Tsimberidou AM, Eggermont AMM, Schilsky RL (2014) Precision Cancer Medicine: The Future Is Now, Only Better. Am Soc Clin Oncol Educ B 34:61–69. https://doi.org/10.14694/edbook_am.2014.34.61
76. Bossart J, Tugrul TK, Kurt S (2014) Fixed-Dose Combination Products – What’s in the Clinic? Drug Dev Deliv 14:36–39
77. Mould DR, Upton RN (2012) Basic concepts in population modeling, simulation, and model-based drug development. CPT Pharmacometrics Syst Pharmacol 1:1–14. https://doi.org/10.1038/psp.2012.4
78. Rajman I (2008) PK/PD modelling and simulations: utility in drug development. Drug Discov Today 13:341–346. https://doi.org/10.1016/j.drudis.2008.01.003
79. Tuntland T, Ethell B, Kosaka T, et al (2014) Implementation of pharmacokinetic and pharmacodynamic strategies in early research phases of drug discovery and development at novartis institute of biomedical research. Front Pharmacol 5 JUL:1–16. https://doi.org/10.3389/fphar.2014.00174
80. Upton RN, Mould DR (2014) Basic concepts in population modeling, simulation, and model-based drug development: Part 3-introduction to pharmacodynamic modeling methods. CPT Pharmacometrics Syst Pharmacol 3:1–16. https://doi.org/10.1038/psp.2013.71
81. Dayneka NL, Garg V, Jusko WJ (1993) Comparison of four basic models of indirect pharmacodynamic responses. J Pharmacokinet Biopharm 21:457–478. https://doi.org/10.1007/BF01061691
82. Mould DR, Upton RN (2013) Basic Concepts in Population Modeling, Simulation, and Model-Based Drug Development—Part 2: Introduction to Pharmacokinetic Modeling Methods. CPT Pharmacometrics Syst Pharmacol 2:e38. https://doi.org/10.1038/psp.2013.14
83. Karlsson MO, Savic RM (2007) Diagnosing model diagnostics. Clin Pharmacol Ther 82:17–20. https://doi.org/10.1038/sj.clpt.6100241
84. Overgaard R V., Ingwersen SH, Tornøe CW (2015) Establishing Good Practices for Exposure-Response Analysis of Clinical Endpoints in Drug Development. CPT Pharmacometrics Syst Pharmacol 4:565–575. https://doi.org/10.1002/psp4.12015
85. Zhao L, Hongshan L, Marathe A, et al (2016) New Advancements in Exposure-Response Analysis to Inform Regulatory Decision Making BT - Pharmacokinetics in Drug Development:
P a g e | 55
Problems and Challenges in Oncology, Volume 4. In: Bonate PL, Howard DR (eds). Springer International Publishing, Cham, pp 303–317
86. Zhu H, Wang Y (2011) Evaluation of false positive rate based on exposure-response analyses for two compounds in fixed-dose combination products. J Pharmacokinet Pharmacodyn 38:671–696. https://doi.org/10.1007/s10928-011-9214-4
87. Pinheiro J, Duffull S (2009) Exposure response - getting the dose right. Pharm Stat 8:173–175. https://doi.org/10.1002/pst.401
88. Minasian L, Rosen O, Auclair D, et al (2014) Optimizing Dosing of Oncology Drugs. Clin Pharmacol Ther 96:572–579. https://doi.org/10.1038/clpt.2014.153
89. Wang Y, Bhattaram AV, Jadhav PR, et al (2008) Leveraging Prior Quantitative Knowledge to Guide Drug Development Decisions and Regulatory Science Recommendations: Impact of FDA Pharmacometrics During 2004-2006. J Clin Pharmacol 48:146–156. https://doi.org/10.1177/0091270007311111
90. Bhattaram VA, Bonapace C, Chilukuri DM, et al (2007) Impact of Pharmacometric Reviews on New Drug Approval and Labeling Decisions—a Survey of 31 New Drug Applications Submitted Between 2005 and 2006. Clin Pharmacol Ther 81:213–221. https://doi.org/10.1038/sj.clpt.6100051
91. Greco WR, Bravo G, Parsons JC (1995) The Search for Synergy: A Critical Review from a Response Surface Perspective. Pharmacol Rev 47:331–385
92. Bliss CI (1939) The toxicity of poisons applied jointly. Ann Appl Biol 26:585–615. https://doi.org/10.1111/j.1744-7348.1939.tb06990.x
93. LOEWE S (1953) The problem of synergism and antagonism of combined drugs. Arzneimittelforschung 3:285–290
94. Lederer S, Dijkstra TMH, Heskes T (2018) Additive dose response models: Explicit formulation and the Loewe Additivity Consistency Condition. Front Pharmacol 9:1–11. https://doi.org/10.3389/fphar.2018.00031
95. Rocchetti M, Del Bene F, Germani M, et al (2009) Testing additivity of anticancer agents in pre-clinical studies: A PK/PD modelling approach. Eur J Cancer 45:3336–3346. https://doi.org/10.1016/j.ejca.2009.09.025
96. Frances N, Claret L, Bruno R, Iliadis A (2011) Tumor growth modeling from clinical trials reveals synergistic anticancer effect of the capecitabine and docetaxel combination in metastatic breast cancer. Cancer Chemother Pharmacol 68:1413–1419. https://doi.org/10.1007/s00280-011-1628-6
97. Goteti K, Garner CE, Utley L, et al (2010) Preclinical pharmacokinetic/pharmacodynamic models to predict synergistic effects of co-administered anti-cancer agents. Cancer Chemother Pharmacol 66:245–254. https://doi.org/10.1007/s00280-009-1153-z
98. Koch G, Walz A, Lahu G, Schropp J (2009) Modeling of tumor growth and anticancer effects of combination therapy. J Pharmacokinet Pharmacodyn 36:179–197. https://doi.org/10.1007/s10928-009-9117-9
P a g e | 56
99. Wicha SG, Chen C, Clewe O, Simonsson USH (2017) A general pharmacodynamic interaction model identifies perpetrators and victims in drug interactions. Nat Commun 8:1–11. https://doi.org/10.1038/s41467-017-01929-y
100. Kwon KC, Lee C (2017) Analysis of Fixed-Dose Combination Products Approved by the US Food and Drug Administration, 2010-2015: Implications for Designing a Regulatory Shortcut to New Drug Application. Ther Innov Regul Sci 51:111–117. https://doi.org/10.1177/2168479016663263
101. Agrawal P (2015) Advantages and Challenges in Drug Re-Profiling. J Pharmacovigil s2:2–3. https://doi.org/10.4172/2329-6887.S2-e002
102. Dugger SA, Platt A, Goldstein DB (2017) Drug development in the era of precision medicine. Nat Rev Drug Discov 17:183–196. https://doi.org/10.1038/nrd.2017.226
103. Mathur S, Sutton J (2017) Personalized medicine could transform healthcare (Review). Biomed Reports 7:3–5. https://doi.org/10.3892/br.2017.922
104. Pierrillas PB, Fouliard S, Chenel M, et al (2018) Model-Based Adaptive Optimal Design (MBAOD) Improves Combination Dose Finding Designs: an Example in Oncology. AAPS J 20:39. https://doi.org/10.1208/s12248-018-0206-9
105. Nøhr-nielsen A, Lange T, Forman JL, et al (2020) Demonstrating Contribution of Components of Fixed-Dose Drug Combinations Through Longitudinal Exposure-Response Analysis. AAPS J 2020 222 22:1–14. https://doi.org/10.1208/s12248-020-0414-y
106. Marshall S, Burghaus R, Cosson V, et al (2016) Good Practices in Model-Informed Drug Discovery and Development: Practice, Application, and Documentation. CPT Pharmacometrics Syst Pharmacol 5:93–122. https://doi.org/10.1002/psp4.12049
107. EMA/CHMP (2009) Guideline on Clinical Development of Fixed Combination Medicinal Products. In: CHMP/EWP/240/95 Rev. 1. https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-clinical-development-fixed-combination-medicinal-products-revision-1_en.pdf. Accessed 28 Nov 2019
108. Standing JF (2017) Understanding and applying pharmacometric modelling and simulation in clinical practice and research. Br J Clin Pharmacol 83:247–254. https://doi.org/10.1111/bcp.13119
109. Maloney A (2017) A New Paradigm. “Learn - Learn More”; Dose-Exposure-Response at the Center of Drug Development and Regulatory Approval. Clin Pharmacol Ther 102:942–950. https://doi.org/10.1002/cpt.710
110. Riggs MM, Seman LJ, Staab A, et al (2014) Exposure-response modelling for empagliflozin, a sodium glucose cotransporter 2 (SGLT2) inhibitor, in patients with type 2 diabetes. Br J Clin Pharmacol 78:1407–1418. https://doi.org/10.1111/bcp.12453
111. Retlich S, Duval V, Graefe-Mody U, et al (2015) Population Pharmacokinetics and Pharmacodynamics of Linagliptin in Patients with Type 2 Diabetes Mellitus. Clin Pharmacokinet 54:737–750. https://doi.org/10.1007/s40262-014-0232-4
112. CHMP (2016) Glyxambi: EPAR - public assessment report.
P a g e | 57
http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Public_assessment_report/human/003833/WC500216974.pdf. Accessed 8 Dec 2019
113. R Core Team (2016) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.
114. Baron KT (2018) mrgsolve: Simulate from ODE-Based Population PK/PD and Systems Pharmacology Models.
115. Bates D, Mächler M, Bolker B, Walker S (2015) Fitting Linear Mixed-Effects Models Using lme4. J Stat Softw 67:. https://doi.org/10.18637/jss.v067.i01
116. Jusko WJ, Ko HC (1994) Physiologic indirect response models characterize diverse types of pharmacodynamic effects. Clin Pharmacol Ther 56:406–419. https://doi.org/10.1038/clpt.1994.155
117. Jordan P, Gieschke R (2005) Explicit solutions for a class of indirect pharmacodynamic response models. Comput Methods Programs Biomed 77:91–97. https://doi.org/10.1016/j.cmpb.2004.02.002
118. Hu C, Zhou H, Sharma A (2017) Landmark and longitudinal exposure–response analyses in drug development. J Pharmacokinet Pharmacodyn 44:503–507. https://doi.org/10.1007/s10928-017-9534-0
119. Moore TJ, Zhang H, Anderson G, Alexander GC (2018) Estimated Costs of Pivotal Trials for Novel Therapeutic Agents Approved by the US Food and Drug Administration, 2015-2016. JAMA Intern Med 178:1451–1457. https://doi.org/10.1001/jamainternmed.2018.3931
120. Tran E (2017) Fixed-Ratio Combinations. Clin Diabetes 35:242–246. https://doi.org/10.2337/cd17-0037
121. Santonja A, Sánchez-Muñoz A, Lluch A, et al (2018) Triple negative breast cancer subtypes and pathologic complete response rate to neoadjuvant chemotherapy. Oncotarget 9:26406–26416. https://doi.org/10.18632/oncotarget.25413
122. Nielsen DL, Palshof J, Brünner N, et al (2017) Implications of ABCG2 Expression on Irinotecan Treatment of Colorectal Cancer Patients: A Review. Int J Mol Sci 18:1926. https://doi.org/10.3390/ijms18091926
123. Taylor NMI, Manolaridis I, Jackson SM, et al (2017) Structure of the human multidrug transporter ABCG2. Nature 546:504–509. https://doi.org/10.1038/nature22345
124. A/S SO Investigation of SCO-101 in Combination With FOLFIRI for Patients With Metastatic Colorectal Cancer (mCRC) With Acquired Resistance to FOLFIRI - Full Text View - ClinicalTrials.gov. https://clinicaltrials.gov/ct2/show/NCT04247256. Accessed 2 Apr 2020
125. Wang X, Zhang H, Chen X (2019) Drug resistance and combating drug resistance in cancer. Cancer Drug Resist. https://doi.org/10.20517/cdr.2019.10
126. Nøhr-Nielsen A, Bagger SO, Brünner N, et al (2020) Pharmacodynamic modelling reveals synergistic interaction between docetaxel and SCO-101 in a docetaxel-resistant triple negative breast cancer cell line. Eur J Pharm Sci 105315. https://doi.org/10.1016/j.ejps.2020.105315
P a g e | 58
127. Dey G, Bharti R, Das AK, et al (2017) Resensitization of Akt Induced Docetaxel Resistance in Breast Cancer by “Iturin A” a Lipopeptide Molecule from Marine Bacteria Bacillus megaterium. Sci Rep 7:17324. https://doi.org/10.1038/s41598-017-17652-z
128. Bagger SO, Drejer J, Brünner N, et al (2018) Abstract A144: Sensitization of docetaxel-resistant breast cancer cells to docetaxel by the VRAC modulator SCO-101. American Association for Cancer Research (AACR), pp A144–A144
129. Fr Brunsvig P, Andersen A, Aamdal S, et al (2007) Pharmacokinetic analysis of two different docetaxel dose levels in patients with non-small cell lung cancer treated with docetaxel as monotherapy or with concurrent radiotherapy. BMC Cancer 7:197. https://doi.org/10.1186/1471-2407-7-197
130. Cho HJ, Kim JE, Kim DD, Yoon IS (2014) In vitro-in vivo extrapolation (IVIVE) for predicting human intestinal absorption and first-pass elimination of drugs: Principles and applications. Drug Dev Ind Pharm 40:989–998. https://doi.org/10.3109/03639045.2013.831439
131. Darwich AS, Ogungbenr K, Hatley OJ, Rostami-Hodjegan A (2017) Role of pharmacokinetic modeling and simulation in precision dosing of anticancer drugs. Transl Cancer Res 6:S1512–S1529. https://doi.org/10.21037/tcr.2017.09.14
P a g e | 59
Papers